Paul Anthony Samuelson-Foundations of Economic Analysis-Cambridge, Harvard University Press, (1947)

HARVARD ECONOMIC STUDIES
VOLUME LXXX
AVARDED THE DAVID A. WELLS PRIZE FOR THE YEAR 1941-42 AND
PUBLISHED FROM THE INCOME OF THE DAVID A. WELLS FUND. THIS
PRIZE IS OFFERED ANNUALLY, IN A COMPETITION OPEN TO SENIORS
OF HARVARD COLLEGE AND GRADUATES OF ANY DEPARTMENT OF
HARVARD UNIVERSITY OF NOT MORE THAN THREE YEARS. STANDING,
FOR THE BEST ESSAY IN CERTAIN SPECIFIED FIELDS OF ECONOMICS
THE STUDIES IN THIS SERIES ARE PUBLISHED BY THE DEPARTMENT OF

ECONOMICS OF HARVARD UNIVERSITY, WHICH, HOWEVER, ASSUMES NO
RESPONSIBILITY FOR THE VIEWS EXPRESSED
FOUNDATIONS OF
ECONOMIC ANALYSIS
BY
PAULANTON _SAMUSN
PROFESSOR OF ECONOMIC-S.: "" _ ,. ,
'iASSAc'HU's' rra ' 05TECHNOLOGY
@@
@;
Mathematics is :: Language
J. WILLARDGrans
UNIVERSITY OF IODHPUR UBRARE
CAMBRIDGE
HARVARD UNIVERSITY PRESS
Cowman, {947
BY THF PRESIDENT AND FELLOWS OF HARVARD COU-EGP
NINTH PRINTING. I971
DISTIIBUTED i GREAT BRITAH DY

OXFOID UNIVERSITY PRFSS
WNDON
UNIVERSITY F Igor-twa I_JBRARE

ACC. NO.!uuuoo-qucSUB-oooml
h L. __
CALL NO -; - - M
sun 674-33093
PRINTED IN rm-: UNITED auras or AMERICA
To
MY PARENTS
PREFACE
1'r-IE ORIGINAL version of tl1is book sub111itted to the David A.
vVells Prize Committee of Harvard University in 1941 carried the
subtitle, "The Operational Significance of Economie Theory.'' At
that time most of the material presented was already several years
old, having been conceived and \vritten primarily in 1937. Further
delay in publication 11as been necessary because of the \var, and
because of the addition of supplementary treatise-Iike n1aterial
going beyond the original conception of the work as indicated by
its subtitle.
Because of the pressure of war work I have not been able to
do full justice to the literature of the last few years, nor even to
include all of the developments of n1y own thinking. Fortunately,
the passage of time has dealt kindly with the analysis contained
here, and where it abuts upon the topics treated in Professor
Hicks's masterly Valite and Capital, the similarity in point of view
11as been reassuring.
My greatest debt is to Marion Crawford Samuelson whose con-
tributions have been ail too many. The result has been a vast
mathematical, economic, and stylistic improvement. Without her
collaboration the book \Vould literally not have been \vritten, and
no perfunctory uxorial acknowledgment can do justice to her aid.
Nor can tl1e quaint modem custom of excluding the value of a \vife's
services fron1 tl1e national income condone her exclusion from the
title page.
My thanks for prolonged stimulation 0\1er rr1any years must go
out to Professors Schumpeter, Leontief, and E. B. Wilson, while
each of a legion of Harvard graduate students has left his mark
upon what follows. The reader will note my dependence upon the
sterling contribution to Welfare Economies of Professor Abran1
Bergson. Grateful acknowledg1nent is made to the Social Science
Research Council and to the Society of Fellows of Harvard Uni-
versity for the opportunities they provided for pursuit of inde-
pendent research, and to the Department of Economies of Harvard
University for their courteous acceptance of the wartime delays
in publication.
vu
vm PRE FACE
Acknowledgment IS made to the editors of Economemca and

the Rewe'wof Fconamzc Stanstzcs for permxssmn to reproduce parts
of my prevxouslypublished articles Chapters IX and X are taken
almost entirely from two artncles that appeared m Econometma
while part of Chapter XI appeared m the Remew of Economzc
Statzstzcs
P A S
CAMBRIDGE MASSACHUSETTS
JANUARY 1945
BOOK REVIEWS 497
out ofthirty-oddnomogramscollectedin the Appendixand whichin the
wordsofthe author"mayproveusefulin the fieldsofengineering, produc-
tion,business,and statistics"thereare onlytwo taken fromthe fieldof
Thesearea nomogram
statistics. forthestandarddeviation(p. 170).
and a nomogram (p. 171)

coefficient
forthecorrelation
2xy
r=-:
V2x2 \/y2
Both of thesenomograms can be used onlyafter2x2, 2y2, 2xy have been

calculated.Sincethe computation of 2x2, 2y2, 2xy is by farthe mostlabo-
riouspart,it seemsa littlenaive to expectthat the above nomograms will
introduce savingoftime.
a significant
FoundationsofEconomicAnalysis.Paul AnthonySamuelson(Professorof Eco-

nomics, Massachusetts Instituteof Technology,Cambridge,Mass.). Harvard
Economic Studies, Vol. 80. Cambridge38, Mass.: Harvard UniversityPress
(38 QuincyStreet),1947. Pp. xii,447. $7.50.
REVIEW BY GERHARD TINTNER
ResearchAssociate,Departmentof Applied Economics
CambridgeUniversity,Cambridge,England
ON LEAVE:Professorof Economicsand Mathematics
Iowa State College,Ames, Iowa
HIS is a very important contribution in the fieldof pure economicswhich

I oughtto be ofinterestto econometriciansand those statisticianswho con-
cern themselveswith economic matters.
The author endeavors throughoutto derive 'operationally meaningful"
propositions in economics from the assumption of maximizing behavior.
Operationally meaningfulstatements are hypothesesabout empirical data
which could conceivably be refuted,if only under ideal conditions. These
ideas are related to those of some modernpositivistphilosophers,especially
Bridgeman. It is perhaps unfortunatethat the author has not included a
methodologicalchapterin his book. An explicitstatementofhis philosophical
ideas may have contributedto the understandingof his book. It may have
helped some economistswho are sceptical about the applicability of radical
positivismand behaviorismin the social sciences.
This content downloaded from 192.12.88.143 on Wed, 3 Sep 2014 08:19:34 AM

All use subject to JSTOR Terms and Conditions
498 AMERICAN STATISTICAL ASSOCIATION JOURNAL SEPTEMBER 1948
The firstpartofthe bookis concernedwitheconomicstatics.It also in-

cludesa chapteron welfareeconomics,whichseemsa littleout of place.
The theoryofmaximain all itsramifications is treatedat greatlength.There
is also a moresystematicmathematical appendixonthissubject.
On thebasisofa verygeneralapproachthetheoryofcostand production
and also the theoryof consumerbehavioris ably stated.An interesting
featurein thetreatment is thediscussionofmaximaifthefunctions in ques-
tionare not continuous. This is a veryimportant advance,thoughthe dis-
continuities consideredmay not be the mostimportantones met within
economiclife.
Monopolyand monopolistic competition get verylittleattention, to say
nothingofmorecomplicated situationslikeduopolyand bilateralmonopoly.
This narrowness ofthepointofviewmakesthebooka littlelessusefulthan
it mighthavebeen.
A shortsectionon the economictheoryof indexnumbers(pp. 146-156)
shouldbe of particularinterestto the statistician.It is probablythe best
shortstatementavailable about the moderntheoryof cost of livingindex
numbers.
A veryinteresting chapteronwelfare economicsconcludesthisfirst section
ofthebook.In itsbriefcompassit containsmostoftheimportant newresults
ofthemoderntheoryinthisfield.
The secondpartofthebook deals witheconomicdynamics.But thedis-
cussionis almostentirely baseduponpurelyformalconsideration ofequilib-
riaand theirrelationto comparative statics.The "correspondence" principle
of the authorstatesthat "the problemof stabilityof equilibrium is inti-
matelytiedup withtheproblemofderiving fruitfultheorems in comparative
statics"(p. 258). The authorsucceedsin reachingveryinteresting conclu-
sionsin thisway.His workis inspiredby themoderntheoryofdynamics(in
physics),especiallyby theworkofGeorgeD. Birkhoff (DynamicalSystems,
1927). A mathematical appendixdeals withthe theoryof difference and
otherfunctional equations.The examplesgivenin thissectionof the book
includealso a veryinteresting "dynamization" of the Keynesiansystem.
But it appearshereand elsewhere thatnotmanydefinite conclusions can be
drawn,evenaftermakinga numberofsomewhat arbitraryassumptions.
The secondpartofthebookis hencesomewhatdisappointing to theecon-
omist.It is evidentthatnot manyvalid and interesting conclusions can be
drawnfromsuch narrowassumptionsas made by the author.The whole
theoryofanticipations and expectations is eitherneglectedor appearsonly
incidentally. Uncertainty hardlygetsan adequate treatment in the small
spaceallottedto it. Samuelson'sbookis in thisrespectinferior to theearlier
workof J. R. Hicks (Value and Capital,1939) and otherswho have en-
deavoredto deal withproblemsof economicriskand uncertainty. I believe
that especiallythe theoryof formation of anticipations formsa truelink
betweenstatic economicsand the moreusefuland interesting economic
dynamics.

BOOK REVIEWS 499
Chapter11 deals withthe classification of equilibria.This discussionis
again based upon modern(physical)dynamics.The usefulnessof the dis-
tinctionsmade in economicsis not immediately apparent.The subsequent
discussionofthebusinesscycleis veryshortand noton thehighlevelmain-
tainedin theearlierpartsofthebook.
The authordevotestowardstheend ofthebooka fewpagesto thetheory
of stochasticsystems.The treatment is muchtoo briefto yieldimportant
results.Some oftheideas presentedmayhoweverprovestimulating to the
statisticians
working in thisfield,especiallytheindicationsabout nonlinear
stochasticsystems.
In summary, thebookoughtto be recommended veryhighlyto themature
readerwhopossessesan adequateknowledge ofmoderneconomictheoryand
highermathematics. A conclusionwhichhe may drawfromits contentsis
the following:Assumingonly maximizingbehavior (somethinglike the
"economicprinciple"of the elementary text books) veryfewinteresting
conclusionsabout economicbehaviorcan be drawneven by so brillianta
theoreticianas Samuelson.There is a greatdeal of valid and important
economictheorycontainedin hisbook.The readermightbe inspiredby the
comparative failureofpureeconomics to devotehisinterest to a moreprom-
ising approach. Econometricinvestigationscombinemodern economic
theory,whichis so ably presentedin this book, withmodernstatistical
methodsin orderto derivevalidempirical conclusions.

CONTENTS
PART I
I. INTRODUCTION
II. EQUILIBRIUM SYSTEMS AND COMPARATIVE STATICS

Symbolic Formulation
Displacement OfEquilibrium
Illustrative Tax Problem .
Illustrative Market Case .
Summary
III. THE THEORY OF MAXIMIZING BEHAVIOR .
Three Sources Of Meaningful Theorems
A Calculus of Qualitative Relations.
Maximum Conditions of Equilibrium
Displacement Of Equilibrium .
Displacement of Quantity Maximized. .
Auxiliary Constraints and the Generalized Le
Chatelier Principle .
Economic Illustrations
Analysis Of Finite Changes
Analytic Functions
Convertibility into a Maximum Problem
IV. A COMPREHENSIVE RESTATEMENT OF THE THEORY OF
COST AND PRODUCTION.
Statement OfProblems
Conditions of Equilibrium
Secondary Extremum Conditions
Displacement OfEquilibrium.
Boundary or Corner Minima.
Discontnuities 1nthe Production Function
Conditions OfEquilibrium
Determinateness of Equilibrium.
Maximization Of Prot .
Indeterminacy in Purest Competition?
ix
x CONTENTS
Discontinuous Case 80
External Conditions of Equilibrium. 81
Summary . . . . . . 87
V. THE PURE THEORY OF CONSUMER'S BEHAVIOR 90
Evolution of Utility Concept. . 90
Progression1nMathematical Thought. 92
The Demand Functions as a Goal . 96
Conditions of Equilibrium 97
Displacement of Equilibrium. . 100
Meaningful Theorems. . 107
Conclusion . . . 116
A Note on the Demand for Money. 117
Qualications Introduced by Uncertainty . 122
VI. TRANSFORMATIONS, COMPOSITE COMMODITIES, AND RA
TIONING . . . 125
LogarithmicTransformationsand Elasticities.125
General Transformation OfIndependent Variables. 129
Transformation of Dependent Variable. . 133
Transformation
ofPrices... . 135
Demand for a Group of Commodities . . .141
The General Problem of Composite or Aggregate
Commodities .144
The Economic Theory of Index Numbers . . 146
Present Formulations of Index Numbers . 156
Pure Theory of Choice under Rationing . 163
VII. Sons SPECIAL ASPECTS OF 1111-:THEORY or Con
SUMER's BEHAVIOR. . 172
The Cardinal Measure of Utility . . 173
The Assumption of Independent Utilities . . 174
Complementarity. . . 183
Constancy of the Marginal Utility of Income. . 189
Why Consumers Surplus 1sSuperuous . 195
The Many Forms of Consumer's Surplus . . 197
VIII. WELFARE ECONOMICS . . 203
The Social Welfare Function. . 219
Mathematical Analysis . 229
Production Conditions . 230
CONTENTS xi
Pure Exchange Conditions . 236
Interpersonal Optimal Conditions . 243
New Versus Old Welfare Economics . 249
Conclusion . . 252
PART II
IX. THE STABILITY OF EQUILIBRIUM: COMPARATIVESTATICS
AND DYNAMICS . . 257
Introduction . . 257
Comparative Statics . . 258
Stability and Dynamics . . . . 260
The Stability of Multiple Markets . . 269
Analysis of the Keynesian System . . 276
X. THE STABILITY OF EQUILIBRIUM: LINEAR AND NON
LINEAR SYSTEMS . 284
Introduction . 284
Functional Equations and Stationary Solutions . 286
Linear and Nonlinear Systems 288
The Nonlinear Di'erentialEquation InOne Vari
able . . 288
Example: Logistic Law . . 291
The Problem of Higher-Order Stability. . 294
An Example of One-sided Stability-Instability:
Malthusian and Optimum Population Theories. 296
Systems of Equations m n" Variables. 299
The Stability of a Stationary Position which

a Maximum. . ISalso
. . 301
The DifferenceEquation In OneVariable . .302
Analytic Solution . . . . 307
Other Functional Equations. . 308
XI. SOME FUNDAMENTALS OF DYNAMICAL THEORY . 311
Statics and Dynamics. . 311
Causal Systems . . 317
Stationary States and their Generalization. . 320
Resolution of the Problem . 329
Concepts of Stability . . 333
Nature of the Business Cycle . 33S
Endogenous Models . 336
CHAPTER I
INTRODUCTION
The existence of analogies between central features of various theories
implies the existence of a general theory which underlies the particular
theories and unies them with respect to those central features. This
fundamental principle of generalization by abstraction was enunci
ated by the eminent American mathematician E. H. Moore more
than thirty years ago. It is the purpose of the pages that follow
to work out its implications for theoretical and applied economics.
An economist of very keen intuition would perhaps have sus
pected from the beginning that seemingly diverse eldsproduc
tion economics, consumer's behavior, international trade, public
nance, business cycles, income analysispossess striking formal
similarities, and that economy of effort would result from analyzing
these common elements.
I can make no claim to such initial insight. Only after labori
ous work in each of these elds did the realization dawn upon me
that essentially the same inequalities and theorems appeared again
and again, and that I was simply proving the same theorems a
wasteful number of times.
I was aware, of course, that each eld involved interdependent
unknowns determined by presumably efcacious,independent equi
librium conditionsa fact which has always been generally real
ized. But, and this leads me to the second fundamental purpose
of this work, it had not been pointed out to my knowledge that
there exist formally identical meaningful theorems in these elds,
each derived by an essentially analogous method.
This is not surprising since only the smallest fraction of eco
nomic writings, theoretical and applied, has been concerned with
the derivation of operationally meaningful theorems. In part at
least this has been the result of the bad methodological precon
ceptions that economic laws deduced from a priori assumptions
possessed rigor and validity independently of any empirical human
behavior. But only a very few economists have gone so far as
3
INTRODUCTION 5
economists ask the right questions of the facts, but even here I
have some doubts.
In this study I attempt to show that there do exist meaningful
theorems in diverse elds of economic affairs. They are not de
duced from thin air or from a priori propositions of universal truth
and vacuous applicability. They proceed almost wholly from two
types of very general hypotheses. The rst is that the conditions
of equilibrium are equivalent to the maximization (minimization)
of some magnitude. Part I deals with this phase of the subject in
a reasonably exhaustive fashion.
However, when we leave single economic units, the determina
tion of unknowns is found to be unrelated to an extremum position.
In even the simplest business cycle theories there is lacking sym
metry in the conditions of equilibrium so that there is no possibility
of directly reducing the problem to that of a maximum or mini
mum. Instead the dynamical properties of the system are speci
ed, and the hypothesis is made that the system is in stable"
equilibrium or motion. By means of what I have called the Corre
spondence Principle between comparative statics and dynamics,
denite operationally meaningful theorems can be derived from so
simple a hypothesis. One interested only in fruitful statics must
study dynamics.
The empirical validity or fruitfulness of the theorems, of course,
cannot surpass that of the original hypothesis. Moreover, the
stability hypothesis has no teleological 2or normative signicance;
thus, the stable equilibrium might be at fty per cent unemploy
ment. The plausibility of such a stability hypothesis is suggested
by the consideration that positions of unstable equilibrium, even
if they exist, are transient, nonpersistent states, and hence on the
crudest probability calculation would be observed less frequently
than stable states. How many times has the reader seen an egg
standing upon its end? From a formal point of view it is often
convenient to consider the stability of nonstationary motions.
In a good deal of Part II the dynamical behavior of systems is
analyzed for its own sake, regardless of implications for compara
tive statics. And in the last chapters of Part I, I have gone beyond
the original conception of the book to include such subjects as
2L. ]. Henderson, The Order of Nature (Cambridge, Massachusetts: Harvard
__Qgiggsity Press, 1917).
6 FOUNDATIONS OF ECONOMIC ANALYSIS
welfare economlcs Although the logreal content of the theorems
enumerated here 15dxerent there ISan underlymg umty of method
In the beginning 1t\\ as hoped that the d\scusswn could be made
nontechmcal Very quickly It became apparent that such a pro
cedure \xlnle possnblc uould Involve '1 manuscrlpt many tunes
the present smc Moreover I have come to feel that \hrshall s
dictum that nt seems doubtful \\ hether any one Spends his tune
ne]! m reading lengthy translations of economlc doctnnes mto
mathematics that have not been made by lumself should be
exactly reversed The laborious literary orkmg m er of cssen
tlally Simple mathematical concepts such as rs characteristrc of
much of modern economic theory 15not only unrewardmg from
the standpomt of 'ldvancmg the scnce but mvoh es as wellmental
gymnastlcs of '1peculiarly depr'u ed t) pe
On the other hand I have attempted to 'wmd all mathematncal
flounsh and the pure mathematician wall recognwe \ too readily
the essentially elementary character of the tools used My own
Interest ln mathematrcs has been secondary md subsequent to my
Interest m economics Nevertheless the reader may nd some
parts hard gomg To lighten the way l have placed the purely
mathematlcal theorems xn two separate appendrces of \\ thh the
second gives a reasonably self contamed Introduction to the theory
of drfference equations
This in brief is the method of comparalwc statics meaning by

this the investigatton of changes m a system from one posxtionol
equilibrium to another Without regard to the transmonal process
invoh ed in the adjustment By equilibrium is meant here only
the values of variables determined by a set of conditions and no
normative connotation attaches to the term As Willbe shown
later it ts al 'nys p0551bleto set up completely tnvral equilibrium
systems
This method of comparative stattcs is but one Spemalapplies
thfl of the more general practice of sc1entic deduction in which
the behavror of a system (possnbly through time) is dened in
terms of a given set of functional equations and mitial conditions
Thus, a good deal of theoretical phySics conSists of the assumption
of second order differential equations sufCient in number to de
termme the evolution through time of all variables subject to given
initial conditions of posmon and \elomty Similarly in the eld
of economics dynamic systems mvolvmg the relationship between
variables at different pomts of time (e g time derivatit es weighted
integrals lag variables functronals etc) have been suggested for
the purpose of determining the evolution of a set of economicvan
ables through time At a later stage I deal With these dynamic
problems
The concept of an equilibrium system outlined above is ap
phcable as well to the case of a smglc variable as to so called
general ethbrzum involvmg thousands of variables Logically
the determination of output of a given rm under pure competition
ISpreCisely the same as the Simultaneous determination of thous
ands of prices and quantities In every case cetem' panbus as
sumptions must be made The only dterence hes in the fact that
in the general equilibrium analysis of, let us say, VValras the con
tent of the historical disapline of theoretical economics is prac
tartan}; QVbRJLURILThe. things. VJJJXJJ.
are. tales. as. dam. [nr. that
system happen to be matters which economists have traditionally
chosen not to wander as w1thm their provmce Among these
data may be mentioned tastes technology the governmental and
Institutional framework and many others
R Fnsch On the Notion of Equal br um and Drsequ librium Remap of Economtc
's Ill (1936) 100105 ] Tmbergen AnnualSurvey Suggestions onQuantitative
ness Cycle Theory Economemca Ill (1935) 24! 308
EQUILIBRIUM AND COMPARATIVE STATICS 9
It is clear, however, that logically there is nothing fundamental
about the traditional boundaries of economic science. In fact, a
system may be as broad or as narrow as we please depending upon
the purpose at hand; and the data of one system may be the vari
ables of a wider system depending upon expediency. The fruit
fulness of any theory will hinge upon the degree to which factors
relevant to the particular investigation at hand are brought into
sharp focus. And if, for the understanding of the business cycle
a theory of governmental policy is demanded, the economist can
ill afford to neglect this need on the ground that such matters lie
outside his province. As for those who argue that special degrees
of certainty and empirical validity attach to the relations encom
passed within the traditional limits of economic theory, we may
leave to them the task of proving their case.
It is not to be thought that the content of systems as described
above must be restricted to the variables usually considered in
price and value theory. On the contrary, one employs such con
structions throughout the whole eld of theoretical economics in
cluding monetary and business cycle theory, international trade,
etc. It goes without saying that the existence of such systems in
no way hinges upon the employment of symbolic or mathematical
methods. In fact, any sector of economic theory which cannot be
cast into the mold of such a system must be regarded with suspicion
as suffering from haziness.
Within the framework of any system the relationships between
our variables are strictly those of mutual interdependence. It is
sterile and misleading to speak of one variable as causing or de
termining another. Once the conditions of equilibrium are im
posed, all variables are simultaneously determined. Indeed, from
the standpoint of comparative statics equilibrium is not something
which is attained; it is something which, if attained, displays
certain properties.
The only sense in which the use of the term causation is ad
missible is in respect to changes in external data or parameters.
As a gure of speech, it may be said that changes in these cause
changes in the variables of our system. An increase in demand,
i.e., a shift in the demand function due to a change in the data,
tastes, may be said to cause an increased output to be sold. Even
here, when several parameters change simultaneously, it is im
IO FOUNDATIONS 01' ECOAOMIC ANALlSIS
posszble to speak ol'c msataon attnbu table to each except m respect

to limiting rates of change (partial dun atwcs).
Sm nouc lonxwumov
All of the aboxc may be stated compactly m mathematical
form Gwen n \arnbles or unknou ns (x1. , x..) and 17:,greater
or less than n, parameters (a: , a...), e assume n independent
and conmstent lunettonal relationshlps nmoh mg our variables and
parameters These may be written most generally tn :mphcxt
form, each equation umolvmg all \aruables and parameters
f(xu . 1'- 01. an) "= 0.
f(x) . Tn all va...) = on
(1)
"'(xl! val all ! a) = ou

or more compactly,
1.0:], , x. &! an) = 0 (2 = 11 on)
Our equations must not be greater than n m number, for 1fthe)
\\ ere. they could not be both conszstent and Independent, 1!less,
our system \\ l, ungeneral. be untler-deterrmned If our equations
possess certam chamcterzstncs, to be thseussed later, they may be
consndered to determme a unique set of alues of our unknowns
(n, . x.) corresponthng to any prC'lSSlgncdset of parameters
(lot , and)
Tlus functtonal rel'tttonslup m1} be expressed mathematnlly
as follons
1-= {(ah ua...) (t = l, , n) (2)
It should be understood that thxs does not Imply that ne can
express our unknown \anables as any elementary functions of the
parameters (such as polynomial, tngonometnc, or loganthmlc
functions) On the contrary, our conditions of equthbnum m set
(I) mll not m general be expressnbleterms of a mte number of
elementary functions, and even 2!they were, we should not be sure
that they could be solved ecphextlym any Simple terms Ho
ever, thus 15of no particular Importance for the elementary func
are only specnal forms which have been of hustoncal mterest
EQUILIBRIUI AND COJIIPARATIVE STATICS II
in the development of mathematical thought and its physical ap.
plications. If one were to draw free hand either at random or as a
result of a complete set of observations a demand curve relating
price and quantity, this could not in general be more than approxi
mately represented by nite combinations of elementary functions.
And yet it is a perfectly good functional relationship indicating a
certain correspondence between these variables. Furthermore, if
there were any practical need for so doing, it could be tabulated
once and for all, be named, and ever afterwards be accepted as a
member of the family of respectable functions.
It is precisely because theoretical economics does not conne
itself to specic narrow types of functions that it is able to achieve
wide generality in its initial formulation. Still it is not to be for
gotten that the aim of fruitful inference is the explanation of a wide
range of phenomena in terms of simple, restrictive hypotheses.
However, this must be the nal result of our research, and there is
no point in crippling oneself in setting out upon the journey.
If one were omniscient, i.e., if all the implications of any as
sumptions were intuitively obvious, the equations (2) would bc
instantly known as soon as the set (1) were given. Short of such
powers, they could only be solved for to any required degree of
approximation with much effort; but, of course, it could be done,
given the requisite time and patience.
Considering the amount of labor involved, one is tempted to
question the advantage of starting from our equilibrium equations
in (1). Why not begin directly with the equations in (2)? Indeed.
it may be pointed out that these resulting functions between un
knowns and parameters could have arisen from an innity of
possible alternative sets of original equations. In particular, con
sider the set of implicit equations
g(a1,'-,a,n) x, = 0. (1,= 1, --,n) (3)
This can be solved by inspection to give the equations of (2), but
of course the solution is trivial. It is trivial in the sense that the
result is intuitively obvious from the beginning, and not for the
reason that (2) says the same thing as (3). For, after all, this
equivalence is also true for (1) and (2), but their identity is not
trivial in this psychological sense.
It is important not to be confused in these matters, for they lie
12 FOUNDATIONS OF ECONOIIHC ANALISIS
at the foundatlons of scxentzcdeduction. and they have been mls
understood partxcularly often by economists By deductwe rea
somng we are enabled only to reveal to ourselves lmphcatlons
already mcluded our assumptlons We may brmg to exphcxtat
tentton certam formulataons oi our ongmal assumptions wlnch ad
mit of poss1blerefutation (conrmation) by empmcal observatnon
Thrs process may best be regarded as a translatton of our ongt
nal hypothesus mto a different language, but so translatmg
prov1ded, of com-se, that no error m logxchas crept mwe do not
change the nature of our ongmal hypothesrs, neither addmg nor
subtractmg from Its val1d1ty and precxswn
The usefulness of the formulatxon of ethbnum conditions
from wl'uch emerges our solutlon hes un the fact that by so domg
we often gam knowledge concernmg possble and necessary re
sponses of our variables to changes data Without such restric
tlons our theories would be meamngless Merely to state, as was
suggested earlier. that there exnsts a nal functronal relatxonshrp
between all variables and parameters (as of an mnlty of accom
panymg crrcumstances) 1s bare and formal, contammg no hy
pothesns upon the emplrlcal data
It 13 because m a Wlde vanety of cases we can more or less
plausubly assume or hypothesnze certam propertres of our equn
hbrlum equatnons that we are able to deduce u 1th the same degree
of plauslbdlty certain propertles of the resultmg functlons between
our unknowns and parameters For the propertles of the functlons
1n(2) are necessarxlyrelated to the structural characteristlcs of the
equnltbrlum set (1) The properties ordmanly d1scussed m thls
connectlon are not specxcquantltatxve restnctlons upon the {une
trons (as being polynomlals, etc ), but rather consxstof assumptions
wrth respect to slope, curvature, monotomcnty, etc . they are the
land lmplted by the law of dzmmzshmg returns
DISPLACEMENTOF EQUILIanu
It 15easy to show mathematncally how the rates of change of
our unknownc thh respect to any parameter, say on.may be com
puted from our equtllbnum equations As a matter of notation let
__g'(al. . "-)
6 Bar g1(m . .a )
EQUILIBRIUM AND COMPARATIVE STATICS 13
stand for the rate of change of the ith variable with respect to the
parameter al, all other parameters being held constant. It is
necessary because of the ambiguity of the conventional notation
of partial differentiation to make sure which variables are being
held constant.
Such partial derivatives must be evaluated as of a given value
of the set of parameters, and hence as of a corresponding set of
values of our dependent variables. Consider the initial position
(ale) ' ' ' ) amo),
and the corresponding set of unknowns

(x10, ' . 1 xno)
where of course
fi<x10 ' ' ', xnoyalor . ' ' ! amo) = 0: (i =: 1r ' '! n) (4)
and
xio : gi(alor . : amo); (1: = 19 ' ' ': n) (5)
since our unknowns must satisfy the equilibrium conditions.
Dierentiating each equation of (1) with reSpect to al, remember
ing that all other parameters are to be held constant, but that all
our unknowns are variable, we get 2
1 _. 1 _, ... = .. ,
f2) (6&1)
6x1 +fr, ( ad]
6x2 )0 + +f2,1 (336,.
al )0 f l
2 __ 2 __ ... : __ .. a! ,
f1 (a)
6x1 "He (a
6x2 ) + H" 2 (xn
aa, ) _ f 2
. _ . (6)
n(Gal)
f-l'l 2.x."+f22"(ray
aal + +fn "(axnw
al fl
where
i afi(x10, , , , , xno, (11,. . ., @
f : ax,- ,
with all other variables and parameters held constant. Similarly,
,. fi(x1. ' ' ' , xnoy (1109 ' ' ' , amo) .
fa] = 60]
2In matrix terms this is [j,-"][xj/m] = [" fai]
L} FOUNDATIONS OF ECONOMIC ANAL}SIS
Be rt noted tlnt the numernml \alues of these part: 11dematwes
are fully determmed at the (.qmllbnum pomt la question Thus
we have 72linear equatlons wzth constant coefcrents m n unknou ns
[Gin/am)" ext/amp] The solution xalues of these wrllde
pend upon the values of the coeferents thus the partral denva
twes relating our dependent V'mables md parameters are deter
mmed by the structural properttes of our equzhbrrum system
Since (6) represents linear equations their solution for non
srngular cases may be represented m the familiar detenmnantal
form
0 if. A&
(%:) = 'A (le=1 n) (7)
here
j! fll f!
f: fz. fx.
A = If | =
f2 1:. fx.
'md A g 1ndtcates the COflCtOI
of the element of the zth row and the
kth column Or nntruc terms
[ai]
6C1] = UT'Ef-J (8)
ILLusraAnvn TAXPRoan
For concreteness let us apply our mal) 515to two snnple cases
Consxdera rm thh '1gmen demand curve relatmg puce and out
put and '1 gtxen productxon cost schedule relatrng total cost and
output Suppose m addrtlon that the output ol the rm rssubject
to '1tax of t dollars per umt The prot of tl e rm may be wntten
r = tdtzreveme wr'u'lprufrum -ccfrcapvynaw.
= xp(x) C(zx) -h
where
xp(x) = total revenue m function of output
C(x) = lowest total productnon cost at wlnch each output can
be produced
tx = total tax payment
EQUILIBRIUAI AND COMPARATIVE STATICS 15
It is clear that for any given tax rate, say t, the rm will decide
upon some given output to be produced and sold; i.e.,
96"= g(t). (9)
where the functional relationship g corresponds to the functions
in (2). However, we cannot leave the matter in so indenite a
state. We wish to know more than that there exists an equilibrium
output for each tax rate. What is the nature of the dependence
of our variable upon the tax rate regarded as a parameter? Will
an increased unit tax result in a larger or smaller output? It is a
poor theory indeed which will not answer so simple a question.
Let us see if we can arrive at an answer to this through the formula
tion of the conditions of equilibrium.
It is in general assumed that a rm will select that output which
will maximize its net revenue. That is to say, our equilibrium
value for output will emerge as a solution of a simple maximum
problem. Specically, it is necessary 3 for a regular maximum of
prot with respect to x, as of a given tax rate, that
7r(x, t)
x : 0 10
62w(x, t) < 0 ( )
ax2 '
The rst condition merely states that at a maximumthe tangential

slope of the prot function plotted against output must be hori
zontal or equal algebraically to zero. The second condition in
sures that we do not have a minimum.
For the problem at hand our condition of equilibrium may be
derived by simple differentiation to become
;; (mx) - can t = 0, (
where it is assumed for simplicity that the inequality in (10).is
realized for all values of the variable concerned. Now equation
(11) corresponds to our equilibrium set (1), which in thlS case con
tains but one equation due to the fact that we have only the value
of one unknown to determine.
See Mathematical Appendix A, sec. 1.
For each value ol t, there W111

correspond a root of the resultmg
equatron m :: and tlns Willbe our ethbrmm value Thus, the
function (9) may be regarded as the exphcrt SOlllthIlof thlS tmphcut
equatlon
Now what have we gamed by Introducmg the maxrmurn prob
lem for the rm? Does thls enable us to answer our orxglnalques
tron as to the nature of the dependence of the output upon the tax
rate? Let us apply the general method outlmed above to calculate
the rate of change of equlhbnum output wnth respect to the
parameter : Drfferentlatmg (11) With respect to ! we get
%,;ww) C(x)1(%%)= 1. (12)

where
ax
(5?) = $")
In thls Simple case no resort to determmants ISnecessary m order
to achieve a solutxon for
x 1
_ : 2 (13)
(a) ;; [xp(x)com
But this prov1des us thh the answer {orWinchwe have been seek
mg For as a sulcrent conchtlon for a relatwe maxnmum we
know that
$$ Ex1>(x) C<x):| < 0 (14)

Therefore
0
(Es-:5) < 0 or g'(t) < 0 (15)

whrch xswhat mtuxtxon tells us would happen as a result of such a
tax Be rt noted nn passmg that at us assumed that the rm ns
always ethbnum before and after the tax 15ImPOScdand that
the tax affects the cquxlxbrmmonly as mthcated m equation (11)
In any actual case eonsxdcmble attentxon must be devoted to the
problem of vcnfymg the correctness of these assumptlons before
any practical apphcatlon 15made of the conclusuonsreached
EQUILIBRIUM AND COJHPARATIVESTATICS I7
ILLUSTRATIVE MARKET CASEll
Let us consider as another example a market for a good or

service where the price and quantity are determined by the inter
section of hypothetical supply and demand curves. Furthermore,
let us introduce a parameter of shift, a, into our demand curve
(e.g., tastes, tax, rival price, etc.). Here we have two variables,
one parameter, and two equations to dene equilibrium values of
our variables in terms of the parameter. Mathematically,
D(x,a) p = 0,
S(x) _p = o. (16)
As a solution we have
x0 = 1 0),
p0 = "). (
How then will our variables change with changes in a assuming
that an increase in a shifts the demand curve upwards and to the
right? As before, we differentiate our equilibrium relations with
respect to the parameter to get two linear equations,
@@)"
6x a _ (?2)
a _ 3:12.
a
(18)
seia efdo: _o
'
By simple substitution we get
6_Q
(9E7) - ""
6a _ _ _ S"
(19)
5,212
(
a _- _ _
D ,.
- S
Now we know that D/a >. 0 by denition of our parameter of
shift. Thus, (ax/600: 0 depending upon whether 5 D/x.
At rst sight this seems merely to put off the dilemma, to replace
one equation by another. But if we examine the kind of market
This is treated in greater detail in chap. ix below.
l FOUNDATION-S OI ECONOMIC ANAILSIS
under conSIderatton, It wdl appear that the mere fact that the
market 18m stable equllnbrlum m the lmtlal Sltuauon W11!remove
all amb1gu1ty For 1fthe market ISthe famthar Marshalhan con
sumer 5good market stabnhty of equrhbnum by demtlon requnres
that the supply curve cut the demand curve from below (even m
the case of decreasmg cost due to external economles)
Thus,
, D
S > 2935 (20)
Therefore,
61:
(a?) > (21)
Hou ever, the algebralc s1gnof the price change w1llhmge upon
whether there 1sa posntxvelyor negatively mclmed supply curve
For
BP 0 .. , E !
(a) -S (au) (22>
Thus, smce (at/600 ISposmve from (21) (GP/600and 5 must be
of the same Slgn, or
(%;)o s' > o (23)

It ISnnpossnble m the price case to get nd of the nal amblgmty
Suppose, however, that this were a market for a factor of pro
ductlon Here the condltlons for stable equ1hbr1umare famnllarly
dened to be enther a posmvely SlOplngsupply curve mth a nega
tively slopmg demand curve, or nf the supply curve ISnegatwely
mclmed, 1t must be backward rtsmg and steeper than the demand
curve Mathematlcally, our stabnhty condltlons may be expressed
SI
_ < (24:)
Q
x
_ S,
Por thxs case, therefore, the sxgnof the pnce change 15known, whxle
the quantxty change 15'unbxguous dependmg upon the algebra1c
A Marshall Prmczple: of Economus (8th edition) p 346 n l p 806 n I
'} R anks Valueand Capctal (Oxford 1939) chapv
EQUILIBRIU-I AND COMPARATIVE STATICS 19
sign of the slope of the supply curve. ln brief,
6p
(a) > 0,
(6%
& >0. S
Innumerable other examples could be cited. In general, we
should not expect to be able to determine the signs of the rates of
:hange of our variables upon the basis of simple a priori qualitative
'estrictions on our equilibrium equations. This is not because of
:he difculty and complexity of solving a large number of equa
:ions; for these could be solved if sufcient were known about the
)articular empirical values of our conditions of equilibrium. It is
'ather that the restrictions imposed by our hypotheses on our
:quilibrium equations (stability and maximum conditions, etc.)
ire not always sufcient to indicate denite restrictions as to alge
)raic sign of the rates of change of our variables with respect to
my parameter.
Only imagine a change in a parameter which enters into all of
1large number of equilibrium equations causing them simultane
>uslyto shift. The resulting net effect upon our variables could
)nly be calculated as a result of balancing the separate eects
regarded as limiting rates of change), and for this purpose detailed
luantitative values for all the coefcients involved would have
:o be known.
SUMMARY
Before going on in the next section of our work to indicate how

:he economist is enabled to deduce signicant results in a wide
:ategory of cases, it will be well to summarize the thread of the
lrgument thus far.
a. For theoretical purposes an economic system consists of a
lesignated set of unknowns which are constrained as a condition
>fequilibrium to satisfy an equal number of consistent and inde
>endentequations (see equation 1). These are implicitly assumed
;ohold within a certain environment and as of certain data. Some
)arts of these data are introduced as explicit parameters; and, as
1result of our equilibrium conditions, our unknown variables - '
)e expressed in function of these parameters (see equation 2
b The method of camparatwe slam: consnstsof the study of

the reSponses of our equlhbnum unknowns to deSIgnatedchanges
m parameters : e we Wnshto know the properties of the functlons
1n (2) In the absence of complete quant1tat1ve Information con
cermng our equilibrium equatxons xt ushoped to be able to {ormu
late qualitative restnctxons on slopes curvatures etc of our equl
llbnum equatlons so as to be able to derive demte qualltatwe
restr1ct1ons upon the responses of our system to changes m certam
parameters It IS the prxmary purpose of thus work to mdlcate
how thls ISpossnble m a mde range of cconormc problems
CHAPTER III
THE THEORY OF MAXIMIZING BEHAVIOR

THREE SOURCES OF MEANINGFUL THEOREMS
IT WILLbe remembered that in the case of a unit tax on a rm's
output it was possible to state unambiguously the direction of
change of output with respect to a change in the tax rate. This
was so because of the fact that the equilibrium output for each tax
rate was derived from the condition that prot must be at a
maximum.
As will become evident in the course of our exposition, this is
by no means an accidental, isolated case; it is merely an application
of a very general principle of economic method, which lies at the
bottom of much of economic theory. In fact, aside from those
parts of economic doctrine whose results are inconclusiveand the
most ardent advocates of economic theory will admit that this
includes a large part of accepted analysisthere is not much which
cannot be brought under this heading.1
The general method involved may be very simply stated. In
cases where the equilibrium values of our variables can be regarded as
the solutions of an extremum (maximum or minimum) problem, it is
often possible regardless of the number of variables involved to de
termine unambiguously the qualitative behavior of our solution values
in respect to changes of parameters.2
It so happens that in a wide number of economic problems it
is admissible and eVen mandatory to regard our equilibrium equa
tions as maximizing (minimizing) conditions. A large part of en
trepreneurial behavior is directed towards maximization of prots
with certain implications for minimization of expenditure, etc.
1The reader may verify this result by paging through any good economicstextbook,
such as Marshall's Principles, and analyzing the derivation of various theorems stated.
2It may be pointed out that this is essentially the method of thermodynamics,
which can be regarded as a purely deductive science based upon certain postulates
(notably the First and Second Lawsof Thermodynamics). That such abstract reasoning
should in the hands of Gibbs and others lead to fruitful theorems testies to the validity
of the original hypotheses.
21
\Ioreover, 1t ISpossxble to denve operattonally meamngful resta:

two hypotheses on consumers demand functtons from the assum;
tlon that consumers behave so as to manmtze an ordmal prelerenc
scale of quantltles of consumptlon goods and serv1ces (Of com-54
thts does not Imply that they behave ratlonally m any normatn
sense )
It IS not to be thought that m prmcnple all economic result
emerge from these maxxmnzmg assumptlons3 For, as we hav
seen, xt was also possnble to deduce concluswe quahtatlve result
from certaln stablhty assumptlons Nevertheless, many of thes
stabrhty condltlons rest; tmphcntly upon maxxmmng behavror
Moreover. certatn difculties anse here Whlle of course, It 1
always possnble to lay down arbltrary denrtlons of stabxhty, 1
zs 1mposs1ble to deduce them w1thout the lmphcrt Introductxon (
dynamical conSIderatlons concerning the behavnor of a system 01
of stattonary equlltbrlum Depending Upon the dynamxcal set u
enwsaged, dxfferent stablllty condltlons are Imphed Thus gtve
supply adjustments of the type assumed m the cobweb cycle phe
nomenon, xt ISwell known that the ordmary Marshalhan condmo:
of a posmvely nsmg supply curve may not result ln "stable
equxhbnum
It IS true that the ldentlcatlon of equxlnbnum With a stabl
maxtmum posntlon m a sense begs the question of stablhty How
ever, where these extremum condttlons are reahzed It can be show:
that many dynamlc set ups Wlllgnve rxseto dampened oscnllatlon
as a result of small dlsplacements The relatlons between dynam:
theory and the stab1hty of equnhbnum are discussed m the las
chapters
There remams Still another posmbdnty by whlch conclusw.
theorems may be deduced We may know m advance certau
qualxtatwe properttes of our equthbrtum equattons Thus, refer
ence may be made to alleged technologncal and psychologtcal laws
' Thus the denmon of econonuctheory as the study of scarce means thh alterna
uve use: I would regard as too broad from one po nt of VIEW,and much too narro
from another
A stationary equrlnbnum ls stable proVIdmg the specncntxonof mmal condition
differ-mgonly sl ghtly from the statuonary equthbnum values results m an evoluuo
which tends (at last m the lumt) to approach the eqml brtum values See Part I
below
THEORY OF MAXIMIZING BEHAVIOR 23
held to be plausible upon a priori grounds.5 Even here, as many
economists have pointed out, the reasoning may rest in the
last analysis upon certain maximum considerations. Thus, in
demonstrating the validity of the law of diminishing marginal phys
ical productivity we ordinarily point to the fact that rms in pure
competition are in equilibrium under a given set of factor prices.
This would not be possible if the law oi diminishing marginal pro
ductivity did not hold. Again we point to the fact that farmers
do not devote all their attention to raising grain on one square inch
of land; they use much land of the same kind, and even inferior
grades of land. Thus, we really argue backwards from maximizing
economic behavior to the underlying physical data consistent
with it.
There is still another type of problem for which the study of
maximizing behavior is illuminating. In some cases, as we shall
see later, it is possible to formulate our conditions of equilibrium
as those of an extremum problem, even though it is admittedly not
a case of any individuals behaving in a maximizing manner, just
as it is often possible in classical dynamics to express the path of
a particle as one which maximizes (minimizes) some quantity
despite the fact that the particle is obviously not acting consciously
or purposively.
For all of these reasons the study of maximizingbehavior affords
a unied approach to wide areas of current and historical economic
thought. There is, moreover, considerable advantage in discussing
the problem at rst in its full generality. The high degree of
abstractness will be more than compensated for in the ease with
which numerous applications can be deduced as special cases.
A CALCULUS OF QUALITATIVE RELATIONS
Before I enter upon the theory of maximizing behavior, it may
be illuminating to show why intuition and a general feeling for the
direction of things does not carry one far in the analysis of a com
plex many-variable system, such as is characterized by equation
In a problem of any complexity involving a number of variables, intuition is a
poor guide for the reasons discussed in the next section. All presumptions become
dubious. In such cases the economist often' falls prey to the perils of assuming the
equiprohability of the unknown. As a result, every reformulation of the problem results
in changed presumptions. This is no doubt one reason why every terminological
revolution in economic thought brings with it a recasting of belief.
24 FOUNDATIONS OF ECONOMIC ANALlSIS
(1) of the prewous chapter To make matters even slmpler tha:
they usually are m reahty let us suppose that we know the quah
tatnve dnrectron of movement of each of our equllnbnum equation
thh resptct to a change In all of the variables and parameter:
Thus we know at least the algebraic Signof each of the rst partta
denvatnves of the formf, ' orf. It does not matter how ne cam
about such knowledge for example ne mlght condently behev
that ldtVldualS wxll d1v1de an additional dollar of Income lrac
tronally between consumptlon and savmg not m consequence a
some economlc theory of mactmlzatlon but sumply as a result 0
everyday observation
What then can ne sa) concernm the drrectron of change of an'
particular variable m response to a change m some parameter
Accordmg to equatxon (7) of chapter n thts rs gwen by a comph
cated expressron whxch can be wrltten as the quotient of two 12b;
n determmants whose elements are made up of these rst part1a
denvatlves Now m order for our questlon to hate an Immedlat
denite answer we must be able to determine unamblguously th
sngns of each of these determmants From the orlgmal funda
mental denltlon of a determmant each consustsof n' terms eacl
term mvolvmg the product of n elements From our hypothesrza
knowledge we can be sure of the sngnbut not the magmtude of cac]
term Only af all of the n' terms happen to be of the same s:g1
wxll the Slgn of the determinant be unamblguous If our system
1nvolvesten vanables the determmants wxllhave more than three
milllon terms Regarded srmply as a problem m probabthty th
chance that a run of ns length should always haVe one Slgn :
about one out of one With a mxlhonzeros alter nt Therefore unles
some specxal feature lS present we are almost sure to nd It neces
sary to wexgh the magmtude of some terms agamst others whtcl
IS to enter upon quantltatlve problems adnuttmg no answer bj
quahtatwe methods
Nevertheless let us see how far quahtatne methods I!" take
us m the slightly Simpler case where only one parameter ISchanger
m such a manner as to alter only one of the equxhbnum relatlons
As before the sngnsof all rst partxal denvatxves are known Con
s1deratlon of all possnblecases that may anse wrll show that then
15no loss m generahty m assumlng that 1tISthe rst equatlon whtcl'
shnlted that the partial derivative of each 1mphc1tequation wntl
THEORY OF MAXIll/IIZING BEHAVIOR 25
espect to its own variable fx;i is always positive, and that the
:hange in the rst equation with respect to the given parameter is
)ositive. If these conditions are not realized, they can be brought
LbOUtby changes in sign of one or more of the equations or variables
ml the parameter.
By our qualitative knowledge we are able to specify a matrix
if the form
4 :b d: --- d:
:b 4 d: --- a:
. . :|: :l: + -- :1:
Sign f,-' = . . . (1)
:l: i :1: ... +

vhere the n columns represent (x1, - - -, x"), and the n rows repre
ent (fl, . - -, f"). The sign in each element will represent the as
umed known sign of the partial derivative of the variable corre
ponding to that row taken with respect to changes in the variable
:orresponding to that column. By our previous convention, we
lave made all diagonal elements plus. All the remaining elements
canbe of either sign, but in any one case they are assumed to be of
lenite known signs. All in all, there are a vast number of possible
natrices of this kind which could arise in any problem, in fact all
:ogether 2 raised to the n(n 1) power.
We are interested in the direction of change of our unknowns,
if in the signs of (dxl/da, - . -, dxn/da). If nothing is known a
>riori, these may take on any one of all the possible 2" arrange
nents of signs.
+
+
|+|+
++++a 2" (2)
+ _ _ ... _
L_'"'
T-Iowmany of these can now be eliminated as inadmissible on the
)asis of our qualitative knowledge as embodied in the specication
of the matnx (1)? Ideally, we should hope to be able to rule out
all but one of the possxblecombmat1ons so as to get a unique answer
However, xt would be at least desnrable to be able to determine the
Slgn of at least one of our unknowns, or, what comes to the same
thmg to be able to rule out a partlcular half of all the possnble
combmatlons x L/
So much for our asp1rat10ns, turmng to the exact procedure by
whlch we rule out a comblnat1on, we nd ourselves m for dns
appomtment If we substitute mto equat1on (6) of chapter 11the
sngns of the 1nd1cated elements, we shall be able to rule out '1
cornbmatton If and only If lt leads to a contradnctron 1e 1fxtdoes
not add up to zero, or to a negatzve number as It should
Concretely, this can be seen to mean that we can rule out any
one of the above combmatxons of sxgn m (2) Wthh exactly duph
cates any of the last (n 1) rows of (I), or whlch IS the exact
antIthesm of any of these same rows Otherwxse, one of the last
(1: 1) equations of (6) chapter n could not add up to zero as
reqmred by the present hypothesrs In addntton we may rule out
any comb1nat1onof (2) \Vthl'l exactly duphcates the rst row but
we are not able to rule out its antlthems /o 64-5 :_
All told then we are at most able to rule out by quahtatlve
conSIderatlons only (Zn I) combmatlons out of a grand total of
2" posmble comblnatlons Even for moderate srzes of n thusleaves
us w1th all but a mmute fraction of possnbnhtnes And th1s Is the
largest number that we can rule out on such a hypothesns In
many cases we are not able to rule out even thxs many Thus, If
any two rows our orngmal matrrx have exactly the same Signs,
they wxllboth rule out the same combmat1ons Wthllcannot there
fore be added together for fear of double countmg
It can be seen then that purely qualltatxve consxderatronscan
not take us very far as soon as the srmple cases are left behind
Of course If we are w1llmgto make more ngld assumptlons elther
of a quahtatwe or quantltatlve kmd we may be able to Improve
matters somewhat Ordmanly, the economist 15not m possessuon
of exact quantztatne knowledge of the partial der1vat1vesof his
cqunhbnum condntlons None the less 1f he 15 a good applied
Leonomnst, hc may have demte notrons conccrmng the rclatnc
1mportancc of dtffercnt e'ects. the better hrs Judgment m these
matters the better an economlst he Willbe These notions whach
THEORY OF MAXIMIZING BEHA VIOR 27
are anything but a priori in their original derivation, may suggest
to him the advisability of neglecting completely certain effects as
being of a second order of magnitude. In other words, zeros are
inserted into the matrix of (1). In fact, the so-called method of
partial equilibrium consists of nothing more than a liberal sprinkling
of zeros into the equations of general equilibrium. In the hands
of a master practitioner, the method will yield useful results; if not
handled with caution and delicacy, it can easily yield nonsensical
conclusions.
By simple extensions of the above argument we can show how
the presence of a zero in any row permits that row to rule out four
instead of only two combinations. Similarly, r zeros in a row per
mits that row to eliminate 2""l combinations. As before, there
may be duplications in the eliminating inuence of dierent rows.
It is also apparent that denite knowledge concerning the sign of
any one of the variables will enable us to rule out one-half of the
original number of combinations, or 2 combinations. Because
of duplicating eects, denite knowledge of the signs of the changes
in two of the unknowns will enable us to eliminate less than twice
as many as one unknown; actually, by two known signs we can
eliminate 3(2"2) combinations in all. Denite knowledge of the
signs of k unknowns will permit the elimination of all but 2"""
combinations.
It would be possible to illustrate the above calculus of quali
tative relations by reference to a number of well-known economic
problems. Space will permit brief mention of a few. In the
market illustration of the previous chapter there was considered
the case of a simple Marshallian partial equilibrium market in
volving two unknowns, price and quantity, whose equilibrium
values are determined by the intersection of supply and demand
schedules. We may apply our analysis to the determination of the
changes in our variables resulting in an assumed shift of a nega
tively inclined demand curve. If the supply curve is known to be
positively inclined, the sign matrix can be written in the form
[t :1
Page 17.
28 FOUNDATIONS OF ECONOMIC ANALPSIS
ln th1s case, we are able to rule out the maumum number of com
bmatxons, (2n 1), or three m all Smee there are only {our
combmatnons we are left W1tha umque answer, as 15apparent from
the algebraic analyms of the prev10uschapter If the supply curve
IS assumed to be negatlvely Inchned the sngns m the second row
both become plus, and we are able to rule out only the minimum
number possnble,or two We are left then thh a nal amblgmty
wluch cannot be settled except by quantztatwe knowledge or by
means of varlous stabIhty hypotheses
A more lllummatmg and more difcult example nsthat of the
smphed Keynesmn system, desenbed greater detall m chapter
1x \thhout gomg Into detalls here, It can be stated that thls gives
use to a system of three varlables, Interest rate xl, Income x, and
Investment, x; whose equlllbnum values are determmed by three
equatlons, hqmdrty preference. f , margmal efcnencyschedule,f2,
and propenSIty to consume f Makmg the usual assumptlons
concermng the sngns of the Varlous rst order effects, eg that
mvestment vanes mversely wuth the rate of Interest and that the
effect of mterest on consumption 18of an order of magmtude whxch
can be neglected, ue end up w1th a sngnmatm of the form
+ - 0
(Sign fr) = - + (4)
o - +
We may now consxder a shalt m any one of the schedules e g
1 change m the hqmdxty preference schedule brought about by
some polncy measure If we apply the above calculus we are snll
left wnth three posmble combmatlons for (51gn change m x.)
namely, (+ + +), ( + +),or ( ) ln thecase ofa shllt
of the mammal eiuency schedule, we narrow the ch01cedonn to
two, (+ + +) or ( ---) For a Shift m the propensnty to
Lonsume schedule ue end up mth three chonces, (+ + +).
(+ + ).0r(- -_)
lt mll be noted that 1nnone of the cases are no able to make a
demte statement about even one of the variables In the sccfmd
case, Wthh xsthe most favorable, we can only say that an Increase
m the margmal emency schedule wxllenther ranse Interest, mcomc,
and Investment, or else lou cr all three of them Thrs ISa highly
unsansfactory state of affairs partlcuhrly smce lt seems on the
THEORY OF JlIAXIJlIIZING BEHAVIOR 29
face of it doubtful that an increase in the marginal efciency of
capital should lower interest, income, and investment. But it is
only by bringing in the stability considerations of later chapters
that a more denite, and more reasonable deduction can be made.
In the discussion in chapter ix it is shown that stability hypotheses
will leave us with only one combination for the rst case, ( + +);
for the second case, (+ + +) ; and for the third case the two possi
bilities, (+ + +) or (+ + -). This last ambiguity is irremov
able unless still stronger quantitative assumptions are made.
Numerous other economic examples could be mentioned, but
these may be left to the reader.7 Before returning to the problem
of maximizing behavior, I should like to call attention to the fact
that the qualitative calculus is not invariant under transformation
of the variables.
NlAXIMUM CONDITIONS or EQUILIBRIUM

Let us consider a new variable, z, dened by a single-valued
function of our previous variables,
z=f(x1,"',xn,a1,"'yam), (5)
wheref and its partial derivatives of at least second-orderexist and
are continuous in a wide region. If for any preassigned values of
the as, there exists a set of values for the xs, (x1, - - -, x,.), corre
sponding to which z_is at a maximum, then we must have
f(xly ' ":xny alo) "'yamo) f(x10, "'ixnoyaloy "') am0)_ (6)
As a matter of notational convenience this may be written
f(X. a) f(X, 09), (7)
7One last illustration taken from the eld of international trade may be dealt with
briey. Professor Leontief has produced a numeril example illustrating the possibility
that a unilateral payment from one country to another may so shift the terms of trade
in favor of the paying country as to cause it to be better rather than worseoff as a result
of the transfer. Note on the Pure Theory of Transfer," in Explorations in Economics
(New York, 1936), pp. 84-92. The example is relully framed so as to guarantee
indifference curves of the proper curvature for both countries. However, if one sets up
an analytical system, along the lines of the numeril example, one will nd that the
Leantief Eject can only happen for a system in which an increase in demand for a com
modity lowersrather than raises its price. If the latter phenomenon is ruled out as being
anomalous, or incompatible with stability (dened arbitrarily or in terms of a dynamic
set up), then we can by the same action rule out the possibility of the Leantitj' Eject.
where X stands for the arguments (x1, , x), and a stands for
the arguments (ax, , a)
If z represents an absolute maxzmum wrth respect to all ad
m1551blevalues of our Independent variables, then
f(X. c2)< f (X. ) (8)
On the other hand, z0may merely represent a manmum relatwe to
all c's lymg In some restricted nerghborhood of the pomt (X. a)
We know from Mathematlcal Appenduc A, Section II that m
order for z to enjoy a relative maxxmum 1t rs necessary that
6
: 0 = f,,(x1, ; xnopall)! ! an)! (3 = l' ' n) (9)
and
] f: ,,Ohh, 0' (10)

where the h's are arbltrary numbers For a regular relatrve max:
mum the last condrtlon can be wntten
% )f,,,h.h, < o, (11)

not all h's equal to zero In other words, thrs symmetncal homo
geneous quadratic form must be negatrve denite (See Mathe
matlcal Appendix A, Section II )
DISPLACEMENT OF EQUILIBRIUM
The set of equations m (9) can be regarded as our conditions of
ethbrxum correspondmg to the set (1) of the secondchapter, and
may be assumed to yield an explncntsolution for our unknown eqm
hbnum values m terms of the preasszgned parameters
xso = g'(a1, s amo) ( = 11 a 11) (12)
Usmg the method of the prevrous chapter we may easnlysolve
for the rates of change of our solution values wrth respect to the
' From the Implncrt function theorem It 18known that assumlng the demteness of
the quadratic form throughout the whole reg-Ionunder drscussron wallInsure us of the
uniqueness of our relanve mammnm posmon Also our maxamumoondmons are essen
tally Invariant under any non smgular transformation of variables Furthermore an?
properly monotomc functlon of ' enjoys an extremum posntronfor the same values of the
asdoesz
THEORY OF AIAXIMIZING BEHA VIOR 3I
5th parameter by means of the following:
2::fziz,.%;: fxiakO, (i = 1, . . 3.72) (13)

.vhere of course,
a- - (j=1,...,n>
aw.- _ gak'1(a1, : am)- (k = 1, , _ , m) (14)
As in (8) of chapter ii, our solution may be written in the de
:erminantal form
;fziakolf
&. = _' (15)
.vhere
le: 0 ' fr,: 0
II = ' = Ifzijol (16)
far-210 "' fz.zo

md H ;,-is the cofactor corresponding to the element of the ith row
1nd the jth column of the Hessian H. As will appear later, we
ire able in a large class of cases to evaluate the algebraic sign of
:his expression. In what follows where there is no risk of am
)iguity, I shall omit the superscript zero.
First let us derive a relationship of complete generality. Mul
:iply the 15thequatioan

('xi
in (13) by xi/ak respectively to get
kazzj ak =f:ak-CY]:. ('t = 1, . . ., n) (17)
Summing with respect to i over all equations we have
" " ____x.-x-_ " _x._
f 36m. aak "- - % <18>
But from (11)
fzrihihj < 01
l 1
holding for any h's. In particular, for

k-=E ("
32 www TIONSor ECONOMICANALISIS
we get
x x;
or from (18)
" x
fnnaat > (21)
for not all x /ak equal to zero ln words thls compound term
consnstmg of the welghted sum of our unknown rates of change
must demtely be posntne m Sign However thts does not add
very much to our knowledge smce we do not know whnch terms
w1ll be posmve
As mentioned earlier however we are not always Interested ln
parameters wh1chShift each and every one of our ethbnum con
drtxons For thxs would necessrtate a knowledge of the relatne
quantxtatwe importance of each shlft before we could hope to
evaluate the composuze result For thts reason we often mrrow
our problem by consxdermg parameters Wthh cause only one of
our equnhbrtum equatlons to shnlt Let us restnct our attentlon
therefore to a set of such parameters (a; a,.) equal m number
to our unknowns We may number each 111the order correspond
mg to the respectlve equlllbrmm equatlon Wthl! 1t shlfts Then
smce a change unthe kth parameter must leaxe all other equatrons
unchanged we have
fx &= 0 for J ?! k (22)
Now our mequahty of (21) reduces to the more Slmple and more
readxly apphcable condmon
ax
f,... >0 (23)
In words for the class of parameters now under conSIderattonthe
rate of change of the kth vanable thh respect to ltScorresponding
parameter must be of the same sxgn as fml \VlllCll
m turn 13posxtwe
nf the Shift of the ethbnum equatlon 15m the dlrectxonof an
merease m x,,
Thls can be vened by the computatlon mdxcated (15)
Accordmg to our present hypothes1s
1332= _fz.a,Hu (24)
am. }!
In Mathematical Appendix A, Section III, it is shown that
11k]:
<0, (k=1,-'-,n) (25)
as a condition for a true maximum. Hence,
}: __ HH:
era, 6a);" (fzkak)2
(T ), (26)
or
x,
f, 67; > 0. (2, )
Let us examine more closely the nature of our hypothesis em
bodied in (22) that each parameter shifts but one condition of
equilibrium, leaving all others unchanged. In the rst place this
does not mean that a change in the ith parameter results in a
change in the ith variable alone. On the contrary, a change in
any parameter will typically result in a change in all variables.
Our hypothesis merely says that this must come about through a
shift in but one schedule with movements along the remaining
schedules.
As I shall later argue, the assumption of this hypothesis does
not involve any serious loss of generality and still includes the vast
majority (in fact, it is hard to nd exceptions) of relationships
contained in current economic theory.
The most general function for which the partial differential
equations of (22) hold may be written
Z = 0(x11' ' ' ! xn) + Bl(xly al) + B2(x2) 0:2)+ ' ' ' + Bn(xm an)' (28)
This may be veried by successive differentiation to obtain

622: (9sz .
= = __ = 0, f 7ek. <29)
ex,-aw, "% ex,-a. Or J
For the rest of this chapter unless otherwise indicated I shall

consider functions of this restricted type. The inequality of (21)
still holds in any case, but is less immediately applicable. After
all, generality is not an end in itself. A theory may be so general
as to be useless. It is for simple theories which have wide appli
cability that we must look.
DISPLACEMENT OF QUANTITY MAXIMIZED

We have Seen how the ethbnum quantntnes (x, x,.j
change when parameters vary Any functlon of them also vanes
m a determmable way In partlcular the quantlty to be 1113le
rzed z wnll change and tts laws of change take on a very snmple
form Let
z = f(x, x,. a) (30)
where 0:nsany parameter Let the x 5 be functions of a determined
by a mammum posmon 1e by
53:=f(x. x,. a) = 0 (z = 1 n) (31)
Ther dz " dx a
36::sz d"; +f.==0+fa=a: (32)
That 15 the rst order change m 2 ISexactly equal to the change
z when the x 8 are not varymg optimally so as to keep z at a
maXImum only to terms of a higher order 15there a dlfference 111
the way 2 as changing A snmlar relanon can be shown to hold
m the case of a constralned maxnnum
This 15the famlhar relatxon of tangency between the envelope
of a family of curves and the curves whzch1ttouches It has many
economlc applicatlons of whtch only a few need be mentioned In
the famous dlspute between Professor Vmer and lns draftsman
Dr Wong the question arose as to the correct relatlonshnpbetween
long and short run curves The short run curve ISdrawn up by
mtmmrzmg total (and average) costs for each and every output a:
of gnen amounts of somexed factor The long run curve requires
lowest total (and umt) costs for every output as plant xsoptxmay
adJusted But accordmg to our theorem a change m the parameter
output wrll to the rst order result m the same change total (and
mm) mais mrerr pn'anrrs
xed aswhen pxcmtas @ma'djaw
Dr Wong was right then msnstmg upon tangency
Another example rs provxded by Professor V1ner' Let labor
marginal cost be the rate of Increase of total costs when only labor
' ] V ner Cost Curves and Supply Curves Ze Lcchffur Nalxmalokonom III
(1932) 2346
" ] V ner Stud :: m the Theory of In amat ouai Trade (New York Harper 1937)
516
is varying so as to produce the extra output. It is to be dis
tinguished from marginal cost of labor which appears in monopsony
theory, and the marginal labor cost which means the increase in
one component of costs as all factors are Optimally varying. Let
marginal cost be the rate of increase of total costs with respect to
output as all factors vary optimally. Then our theorem states
that labor-marginal cost equals marginal cost equals any other
factor-marginal cost. As Professor Viner has pointed out with
great insight, at the margin (i.e., neglecting differential coefcients
of higher-orders) all factors are perfectly indifferent substitutes.
The classical economists, lacking the precise notion of an inni
tesimal, were forced to employ the concept of a broad extensive
margin. (Viz., Ricardos no-rent land, and J. B. Clarks famous
zone of indifference.)
For nite movements, however small, higher-order terms will
make the change in the quantity maximized (minimized) different
when the unknowns are optimally adjusted from the change when
they are held constant. Actually,
d2z n d2xi n dx.- d(fi) n dx,
E5= Zfi+
1 a lg--l' ailing-Ff (33)
dx,
0 + O+ (if-{a- +fai
since f.; = O. .
The higher-order change in 2 when all x's are held constant lS
given by
2
%=f- @@
The difference between the former change and the latter is as
follows:
m$Mm>
dzz
2z " d_x,-
()35
because of equation (21) above. . _.
If the changing parameter affects prots or ordinal utility ad
versely, it does so less when output and consumption are optimally
adjusted to the new circumstances. If the parameter improves
uFor still other examples see Marshalls Principles, Mathematical Apper
Note XIV, pp. 846-852.
30 rUUNv/U'IUNS OF ECONOMIC ANALYSIS
prots (etc) nt does so most when the unknowns are optnmall

adjusted
One could go on to Still hlgher terms These would be seen tl
depend upon the venous partial denvatwes of f and upon dx /da
dx /da:2 etc The changes In z are of hIgher order than the change:
un the x 8 In fact the nth derlvatnve of z depends at most upor
the (n -1)th denvatne of the xs This ns brought out In th:
above equatlons
Auxruamf ConsTRAINTs AND THE GENERALIZLD

LE CHATELIBR PRINCIPLE
If the equnhbrxum of a system ISdetermmed by ewtremum con

dnttons where all unknowns are Independently variable the add:
tron of 'wmhary constramts (satised by the equrhbnum posxtlonj
Wlllleave the equnbnum unchanged If a true relattve maximun
ISatt'uned a movement un an} dtreetlon leads downhlll a fOfthI:
movements along certain subsets of directions wnll be downhlll
The usefulness of what may at rst seem a strange process hes m
the fact that 1t enables us to deduce necessaryeondltlons which thc
equthbnum must satlsfy
Thus wrth plant xed '1 varnble factor w1|l be htred untll Its
wage (marginal cost) equals ltS marginal value producthty In
the long run plant cannot be taken as xed Nevertheless the
short run condltton holds m the long run as well (but not nee
\ersa) smce long run total costs cannot be at a mlmmum unless
short run total costs are as low as possnble Again undiscrzmlnat
mg monopoly the condltton that any gnen output be dwlded
optxmally between two markets (equal margmal revenues) holds
even when total output xs not gwen but xs determmed by cost
con51deratlons
How 15the equlhbrmm displaced when there are no aux1llary
caastramts as cammred to the case whenconstraintsare mpnsed
How does the demand for a factor vary w1th lts pnce when other
factors cannot be optimally adjusted because of time lags etc
Thus type of question IS Important ln thermodynamics as ell as
economlc systems lt admxts of a snmpleans er
Let :: be of the follou mg specual form
2 "' 9(x1 x,.) - (11.1.1-- a rg -- -- aux (36)
uwuny ur MAJMJZINGBEHAVIOR 37
If all unknowns are independently variable, a maximum is deter
mined by the equilibrium conditions
z .
'aT..=6,-(x1,---,xn)ai=0 (7=lv"'in) (37)
and
[H] = [ati]
is the matrix of a negative denite form. How does x.-change when
its conjugate parameter a.-varies?
dx; __ ET,-,
d; - 7:1-< 0. (38)
Let r independent additional linear constraints be imposed such
that
mx,- -x9) = 0, (a = 1, r) (39)

1
where
_ [gl]
ISof rank r.
The introduction of constraints requires that our equ1hbrium
system be modied to assume the following form
0i+r)\ f8- i=01 ('l:=l,",7l)
?, BE a (40)
i g(xi _ xJ'o) : 0! (B 11 ' ': T)

1
where the Xsare Lagrangean undetermined multipliers. _

Let 'H represent the determinant formed by bordering H Withr
rows and columns consisting of the coefcients gf'; 1.e.,
rH= % gil.
0 (B7=1,.,7) (41)
Then the change in x.-with respect to 0:,-when r auxiliary con
straints are imposed, is given by
dx,- _ LH; (42)
(aa.)' % <
if [H] is the matrix of a negative denite form, and the matrlx
% J is negative semi-definite. (Cf. Mathematical Appendix A

Section IV.) Adopting the convention that
H = H,
equation (38) ts in as a special case of (42).
What is the effect upon the rate of change of x. with respect to
ouof adding an additional constraint? Clearly
(a-)
da. ' - (a)
da; ,-1 a
'II IH
"FI 'HiH-r. a+! "
'H 'Hu-i-r.n+r
" 'HH'H'N'r' "+' _ 'H'Hn4-r n+ru (43)
_ 'HrH' '. +v
(' )
= __'_1'J'_'._
'H'H'H". 84?
by a well-known theorem on determinants (Jacobi). __

The denominator is pOSItive because such bordered princ1pal
minors regardless of the number of bordering rows must be of the
same Sign " Hence, the difference is pOSitive
We have the following general theorem.
@) (%)
da. o = da; 1 r-
= 4%)
: dat ul
IM O (44)
While the change in an x with respect to its own parameter is

always negative regardless of the number of constraints, it is most
negative when there are no constraints, only less so when there is
a single constraint, and so forth, until the number of auXiliary
constraints reaches the maximum possible, namely (7: I).13
This explains why economically long-run demands are more
elastic than those in the short-nm. A lengthening of the time
1'See MathematICal Appendix A equation (48) It should be noted that our '
corresponds to ii: there. and r there corresponds to here
" This is a purely mathematical theorem lt corresponds to someof the phenonfucga
which {all
almost under the heading
metaphysical of of
vagueness theitscelebrated prinCiple
formulation. of Le5Chatelier
the latter Becausedo
meaning is often in outbe
t.
and it is used at one and the same time to cover diverse phenomena The above formu
lation explains why the change in volume With respect to a given Change in pressure IS
greater when temperature is constant than when entropy is held constant and temlma
- Permitted to vary in accordance With the conditions of equ'l'bu
period so as to permit new factors to be varied will result in greater
changes in the factor whose price has changed, regardlessOfwhether
the factors permitted to vary are complementary or competitive
with the one whose price has changed.
ECONOMIC ILLUSTRATIONS
It is not hard to point out the relevance of the above for a large
number Of economic problems. First, let us repeat the result of
our previous analysis.
6x,
ft.-a.- aai
> 0. (45)
This states that the direction of the change Of the ith variable
with respect to its corresponding parameter is Ofthe same sign as
f,_.._.(
= B). This quantity may be taken as a criterion. If its
algebraic sign is denite, i.e., unambiguously determinable, then
the sign Ofaxi/aai is also denite.
We have merely to show, therefore, that a wide variety of eco
nomic problems can be so formulated as to yield a conclusive de
termination Ofthe sign of our criterion.
It may be noted that for maximum problems involving a single
variable no restrictions need be placed upon 2 in order that our
criterion be applicable. Let
2 = f(x, a). (46)
Then our criterion becomes
x
f" 5&-> 0. (47)
I list below some random examples chosen to illustrate the
applicability of the criterion.
(a) Let us go back to the example Ofthe second chapter Ofthe
eect of a' unit tax on output. It will be remembered that the
relationship between the equilibrium output and the tax was de
termined to be unambiguously negative. This conclusion can be
quickly Obtained by our present method. Prot is dened as
follows:
1r = an(x,t) = Exx) - C(96)] tx. (48)
Our criterion may be easily computed.
W:: = %: = " I. (49)
40 FOUNDATIONS OF ECONOMIC ANALLSIS
Therefore,
6x
( I) 5 > 0 (sa)
or
x
a <0 (51)
Also. as shox the prcvzous section,
dn' 311'
To a rst approumatron. the change m prots ts unaectedby

the adjustment of output To a more accurate approximation,
prots are reduced less If output nsoptzmally altered
(b) The effects of three other kinds of taxes are also 51mply
derivable Consxder respectwely a percentage tax on gross sales,
a lump sum tax. and a percentage tax on prots The correspond
mg prot functions may be wntten
r = Exx) C(x)] t'xc).
fr = Exx) C(x)] !" (52)
1r ==Exx) C(x)] - t"'[xp(x) - C(x)J
Our cnterna are respectwely

W:! = _- 5; BPM] < 0.
71', = O, (53)
36.1'"
T:: = ' 'Tr(aTZ : 0
Obvrously. therefore, for the rst case of a tax on gross sales, the
effect of an Increase m the tax rate ISto reduce output The last
nm cases however, present a new feature Our cntenon ISneither
posrtxve nor negatnc, but equal to zero A little reectxonreveals
that our equ111br1ummaxtmum conditions are essentrally unde
pendent of the parameters changed Hence, our equlhbrrum
\alucs remain unchanged, 1e ,
; a 0.
al (54)
x E 0
dl H
THEORY OF MAXHHZING BEHAVIOR 4I
These conclusions are, of course, familiar from the Marshallian
analysis.
(c) Let us now consider a problem which has nothing to do with
taxation, but which gured prominently in the famous cost con
troversy of a few years back. Let us suppose a rm in pure com
petition, i.e., one which can sell as much of its output as it wishes
without affecting the price. Given total cost in function of output,
there will be a determinable output reaction to each given price.
What is the nature of this dependence? From the fact that the
rm is assumed to be in stable equilibrium when it enjoys a proper
relative maximum, we can easily apply our criterion to deduce the
properties of the supply curve. Here
11'= 99(xtp) = px _ C(x)r (55)
1r , _
=p_c@y41 o
Equation (56) can be solved to determine output in function of
price.
x = g(1>) (57)
It is easily veried that
772p E 9911:E 1- (58)
Therefore, d
3% = g(p) > 0. (59)
Even if we relax the requirement that a regular relative maximum
be realized, it still remains true that the supply curve cannot be
negatively inclined. Of course, this does not mean that the mar
ginal cost curve cannot be negatively inclined, but merely that in
such ranges it cannot serve as a supply curve.
(d) It must not be thought, however, that the assumption of
our equilibrium as the solution of a maximum problem is the open
sesame" to the successful unambiguous determination of all possible
questions which we may ask. For it is extremely easy to specify
simple and important problems which cannot be answered even
qualitatively without further knowledge.
Consider the problem of the effect of the introduction of adver
tising expense upon the output of a monopolist. Will an increased
advertising expense result in a larger or smaller output? Here
w=mw=mecmm mm
THEORY OF MAXIJIIIZING BEHAVIOR 43
Under discriminating monopoly all prices are regarded as inde
pendently variable and are adjusted so as to maximize prot.
Prot may be written as
rr= P1D1(1J1)+ P202(P2) "' CEDI + D2(P2)]- (65)
For simplicity we rule out the possibility that all of the demand
at a given price in a given market may not be satised by the
entrepreneur. The removal of this restriction can be easily made.
Here the conditions of equilibrium are
611' , le
=O=DI(P1)+(P1C)v (66)
in; _ _ , dD2
am 0 web:) + (p2 6);;
This results in an optimal set of prices (p10,pg") and quantites
(x10,x2"), and total quantity (X), or (x1+ x20).
In the case of simple monopoly we impose the condition upon
our problem that the price be equal in both markets. Thus,
pl : 12= P: (67)
= 1201(1))+ we) CEM) + Dem]. (68)
gg- = 0 = Diab) + D''(i>)

+ (p 0) (% +%%) (69)
This yields as a solution (12,p), (251
, xz), and (X '). Is it possible
to determine whether
X %-X? (70)
At rst glance there is little resemblance between this and the
examples hitherto discussed. No data have been introduced as
parameters; furthermore, we are not comparing two situations in
denitely close to one another. On the contrary, we seem to be
dealing with two entirely dierent kinds of behavior. The two
solutions seem to result from qualitatively different types of maxi
mum problems. Nevertheless, we may avail ourselves of an arti
ce by means of which we can bring to bear the methods used above.
Let us introduce a parameter k, dened as follows:
&_
p1 _ k. (71)
We may now treat as rndependent varxables (p, h) mstead of
(pl p.) smce there xs a one to one relationship between the two
sets Therefore
n' = F(p1 p) = Fan km) = GU) k) (72)
In the case of dlscrunmatmg monopoly both prices are vanul 1nde
pendently so as to mmmrze prot This IS equivalent to the
condmon that both ?; and !: must be allowed to vary so as to
mamrmze prot Our condttrons of equnhbnum are
2.1-92-0
ap'ap.
~-o
akak (
7
llns must of course determine the same solution as given m (66)

smce e have only transformed our variables
For sunple monopoly xt ls a preass1gned condrtnon of the prob
lem that both pnccs be the same or that k be equal to unity
Thus by contmuous movement of k from umty to {n/pl" we may
pass from srmple to thscrmnnatmg monopoly If then we could
unambrguously state the directxon of the rate of change of output
wnth reSpect to k at each pomt lt would be possrble to determme
whether output would be larger or smaller
In order to ascertam the rate of change of total output With
respect to k let us strll further transform our variables as follows
12:33
PI (74)
X = DUn) + 132(le
Solvmg mversely we get
?! : fl(k X) (75)
m = 10? X)
Thus
1r= HP: P) = FlfUeX) f(k X)] = dk X) (76)
Our equnllbnum condltxon for a gwen le:s
'_<0(kl)_
'_'ax _0 ()77
THEOR 1' OF MAXIMIZING BEHA VIOR 45
According to our criterion
dz
90x1.-dk >
0. (78)
Now the computation of (ml.-,while laborious, is nevertheless

possible. It will be found to depend in a complicated way upon
the curvatures of our demand curves. Its algebraic sign will be
ambiguous, as must therefore be the effect upon total output.
However, anyone who cares to go through the computation can by
so doing determine what additional restrictions must be placed
upon the demand functions in order to insure that output be larger
or smaller.
(f) In later chapters the methods outlined here will be sys
tematically applied to different branches of theory. In order to
illustrate that it is possible to derive unambiguous conclusionseven
with a large number of variables a small part of our later analysis
will be anticipated here.
Let us consider the demand of a rm for a factor of production.
We assume as given a production function embodying the tech
nological relationships between inputs and output, the demand
curve for the nished product, and the prices at which all factors
of production can be bought in unlimited quantities. The condi
tion that prot be at a maximum is sufcient to determine the
value of all unknowns in terms of these data. Thus,
71' total revenue - total factor cost
II R(x) (wlal + 11;)202
+ - - - + wnan), (79)
where w,-and a,-are the respective prices and quantities of the z'th
factor of production. But we are also given a production function
(assumed continuous),
___.x(al . . ., a). (80)
Hence,
7r= f(a,,---,a,,,'w1, ---,w,.)
= R[x(a,, ---, a,,)] -('wlal + + ma). (81)
Recall from (21) our generalized criterion.
z f" " 60
aw.
>0. (82)
Obwously here
fm E 0 (z # J) (83)
Hence,
;
f%:>0 (2:10 on) (84)
But
fan. E ' 1 (z = lv . n) (85)
So
aa.
m<0 (z=1 .n) (86)
This conclusmn holds for any number of factors
The above examples are but a small sample of economlcprob
lems whrch can be regarded as determmed by the solution of
max1ma problems For these, the cntena outlmed above may be
employed to deduce meaningful unamblguous quahtatwe theorems
ANALYSIS OF Fmrra CHANGES
Up untll now the analysm has been conned almost excluswely
to the determination of the algebralc sngnsof Instantaneous rates
of change One cannot leave the matter here, for m the world of
real phenomena all changes are necessarlly mte and Instantaneous
rates of change remam only llmntmgabstractlons It 15Imperatwe
therefore that we develop the implicattons of our analysrs {ormte
changes Fortunately, desplte the Impressroncurrent among many
economists that the calculus can only be applied to mmtesxmal
movements th:s ISeasy done
Let us conSIder for mmplncntya functional relatronsh1p between
one vanable, x, and one parameter, a contmuous and twncedr'er
entlable everywhere on a given Interval
x=g(a) aab (87)
Suppose that we have already ascertamed the fact that the alge
branc srgn of the Instantaneous rate of change of thrs function xs
everywhere negatlve on the dened Interval, 1e ,
g=gi(a)<o aab (88)

It follows then that any mte change m a w1thm thrs Interval wxll
accompamed by a nite change m x m the opposrte sense
For let
Aa = a1 . 0 a b (89)
Ax = x1 - x = g(a') g(a). a a b
We wish to prove that
Aan < 0. Aa 95 O. (90)
By the theorem of the mean
Ax = gel) g<a>= j: g'<a)da = g'Aa. (91)

where
e=a+0(a-a). 0<6<1
But since the derivative g is negative everywhere on the dened
interval, it must be negative for a = e. Therefore, Ax and Aa are
of opposite sign, for
Aan = g'(e)(Aa)2< 0. (92)
This result is intuitively obvious. If, starting from a given
point, a curve is always falling, then one should not expect it to be
higher after a nite distance, continuity being assumed.
The above proof rested on the assumption that the derivative
be unambiguously negative everywhere. This is often an ad
missible hypothesis as we have seen. But what about the cases
in which the derivative is known to be unambiguously negative
only at a point? Thus,
g'(a) < 0. a < a < b (93)
Can anything be said then about nite changes? Our answer is
in the afrmative. It can be shown that for all nite changesin a
smaller than some assignable quantity, there exist correspondingnite
changes in :: of the opposite sense.
For by hypothesis g'(a) is continuous on the given interval.
and of course at a. Hence, from the elementary denition of
continuity there exists a neighborhood around a0 in which g(a)
is always negatiVe, i.e.,
g'(oz)< 0. la - al< h (94)
Thus, from our previous theorem
Aan = (x -x)(a a) < 0. lo: - a|< h (95)
The unphcatron of the last theorem xs Important In order
that the results of all mte movements be unambnguous, It IS
necessary that the Sign of the Instantaneous denvatxve be demte
everywhere For let the denvatxve change sxgn, and xt wxllbe
possnbleto nd contradrctory mte movements
Thus far we have assumed that our equnhbnum equatlons are
such as to enable us to solxe for our unknowns umquely m terms
of our parameters Under what condxtlons wnll thus be posslble>
It w1ll be readxly seen that even where our equllzbnum condltnons
are dened as the result of a maxunum problem there remams the
possnbnhty of multiple relatlve m'mmum posmons Under what
condmons are we able to get a nmque exphcrt solutnon of our
imphcxt equations? What shall be done If multiple solutions are
possnble?
Assume the xmphcxtequattons
f = f,(x1 xa. , a...) = 0, (z = 1, .n) (96)
and a set of values (x.. , x.., (11, , a)satrslymg these equa
tlons It xs known from the Impllcut Functlon Theorem that
there exnsts one and only one expllmt solution of the form
x: = {:(m a...) (t = 1. n) (97)
m a regton around (n x,.,ou, , a)where the followmg
functnonal determinant does not vamsh
len fx. ft.:
fax; jh"! fz'u
I: L !
Smcc we shall be consndermg cluefly the selectxon of regul'ar
maxzmum posmons, we may assume
(- 1)"H = (-1)"lfu,| > 0 (99)
at the pomt of ethbrzum Provndmg thls expressnonremains
posxtzveeverywhere we may be sure of a umque equnhbnum Of
course thxs xssufCIent but not necessary
' See any \dvanccd Calculus
Suppose, however, that the Hessian does change sign a nite
number of times within the region of economically admissible
values. Then the functions in (97) will be multiple-valued with a
nite number of branches. Some of these may be ruled out im
mediately as not constituting maximum positions, namely those
for which
( 1)"H < O. (100)
It is possible to choose from the remaining branches only by re
ferring back to our original maximum problem. Let (x11,- - -, x,.l),
(x12, - -, x,.2), - - - correspond to a preassigned set (al, - - -, a) as
multiple solutions of (96). That set (or sets) is retained for which
f is greatest. This will ordinarily serve to dene our xs as single
valued functions of the a's except at a nite number of points.
At these points it is a matter of indifference which of the possible
alternative solutions is embraced.
Broadly speaking this possibility of multiple equilibrium offers
no serious difculty. For all qualitative results remain. To illus
trate this let us consider the problem in full generality. Let
z = f(xx, -,x
on. an) = 0(x1. ---.xn) + B(x1.a1)
+ 32m, a2) + + B(xan)- (101)
Considera preassigned set of values of our parameters (021,- - -, an),
and a corresponding optimal set of values for our unknowns
(x10,", xn), not assumed to be necessarily unique. Then by
denition of a maximum
j(xl! ' " xny alu, ' ' ', duo) ; f(xloy ' ' ', xnopalor ' ' '! ant!) (102)
where (xl, -, x) take on any values. For brevity this may be

written
f(X, a0) f(X, a) 0. (103)
Consider any other preassigned values of our parameters (ail,
. an), and a corresponding optimal set (x11, ' ' ' , xn1) Then
f(X. a1) f(X". a!) z. 0. (104)
f(Xo! ao) _ f(X1,a0)=_>_0'
Adding both sides, we get
[f(X, a) f(X,a1)] - [f(X1,a) -f(X. 0.0)]2 ' (105)
From the denition of a denite integral this can be written
n
! :0 U8(xll ' " x!!!all! ' ' " an!)
fz(xl; ' ' '- xm aloo ' ' '. bald: 0. (106)
From the same denition this can be further transformed
L: EJ fz,(xh ' ' " xl" al' ' ': an)dxidaj 0- (107)
But from (101)
fac; = Be"; 5 or (i # j)
rm = B 8".' (108)
0' J
? f.:, f .* fx,c(xlv' ' : x!"ab ' " al)dxodai 0. (109)
By the theorem of the mean this may be written
! Kauai a o. (no)
where : is evaluated at an intermediate point. If we consider
movements of the kth parameter alone, others being held constant.
this becomes
Kama, o. (k = 1, --,n) (111)
Hence, if our criterion is denite in sign everywhere, e.g.,
fa'b : BHO: < bn (k = It ' ". 13) (112)
then,
Anga. g 0. (k = I, - - -, n) (113)
This is proof that multiplicity of equilibrium values does not
alter the deniteness of our conclusions with respect to nite
changes. It also shows that the criterion which we applied m
earlier sections to determine the deniteness of instantaneous rates
oi change can be generalized to determine the deniteness of nite
changes. In fact, by a proper limiting process our prevmus
theorems relating to instantaneous rates of change can be deduced
as special cases from this more general analysis. _
The equality sign can be dropped il Aa # 0, and if the-main
mum is proper. It will be noted that in this proof we d1d not
require that 0 be continuous, but merely that the B be continuous
with the derivatives of the required order. This will turn out to
be of considerable importance in the later consideration of discon
tinuities in the production function of a rm.
A very important case is that in which 2 takes the form
z = 0(x1, - . ,x,.) )am. (114)

Then
6(x) 213013303; 9(x) me,, (115)
6(x) ' Z:jagx,- 0(x1) d.-cx}. (116)

Adding and collecting terms, we get
2:30:19 of,-)(xil _- nc.-); 0, (117)

or
AaiAx; ; o. (118)
As illustration of the direct application of our methods to nite

changes, let us revert back to the rm with a given total cost func
tion producing in a purely competitive market. For a preassigned
price, p", the rm will supply a given output, x (not necessarily
uniquely dened, since the rm may be indifferent in its selection
among two or more outputs). Since prot is at a maximum for
this output
[Px C(29)] [px C(x)] g o, (119)
Where9:1may be any output, in particular that output appropriate
to a second price, #, such that
DM C(x)] - [px - C(x)] ; o. (120)
Adding we get
APM = (19- P)(x _ ) % 0, (121)
and 97
Apr > 0 for Ax 75 0, A? ; 0. )
For this analysis it is not necessary that the cost curve be con
tinuous, or be such as to yield a unique optimum output. The
marginal cost curve may be undened at points. have kinks, and
turn up and down many times.
ANALYTIC FUNCTIONS
In cases where z is an analytic function of the :csand the car's,

nite changes can be evaluated by an innite power series in the
Aas, whose coefcients depend upon partial derivatives of all
orders of z at the original position of equilibrium That is,
n o n n 2 o
(>
<->
(i = 1. -- .n) (123)
The general coefcient is oi the form
( dal! "
. . . dan!- )
where
}: m. = n. (124)
l
These coefcients can be computed from-the equilibrium equations

by differentiating as many times as is necessary with respect to
the cars. If the Hessian is not zero, this will yield recursion rela
tions sufcient to get the desired higher derivatives in terms of.the
already computed lower derivatives and the higher partial dcer
tives with respect to the x's.
Convsnnmurv INTO A MAXIMUMPROBLEM
Earlier in this chapter it was indicated that some problems
which do not appear to involve extremum positions can at tunes
be converted into an equivalent maximum or minimum problem.
The advantage to be derived from so doing is merely notational,
since it will require fully as much knowledge to ascertain whether
the conditions of a maximum position are met as would be neces
sary to answer any questions which might be aslted. Furthermore.
there is the danger that unwarranted teleological and normative
welfare signicance will be attributed to a posrtion of equilibrium
" dened. To avoid misunderstanding it is wellto emphastzethat
the conversion of a problem whose economic context does not
suggest any human, purposive, maximizing behavior into a maxi
mum problem is to be regarded as merely a technical device for the
purpose of quickly developing the properties of that equilibrium
position.
Our problem may be stated in the followingway. Given initial
conditions of equilibrium
fi(xlv"'rxmaly1am):O: (7:=11"1n) (125)
such that our unknowns are determined in function of the given
parameters, namely,
x,- = g(a1, ---, a), (i = 1, ---, n) (126)
under what conditions can the set (125) be regarded as the solution
of an extremum problem so that the n indicated equilibrium loci
correspond to the vanishing of the partial derivatives of some func
tion? That is, does there exist a function
Z=f(x1,'-,xa1,---,am) (127)
such that
ft.-(xl: "'yxmaly "'yam) : O; (1'= 1: "',71) (128)
represents the same locus for each value of i as
f"(x1,xa1,-",am) = 0? (i = 1,---,n) (129)
Now the same implicit function may be represented in an in
nity Of ways without changing the locus it denotes.15 It is de
sirable, therefore, that we represent the loci which are our condi
tions of equilibrium in a denite unambiguous form. One such
way is provided by solving explicitly for each variable in turn,
xi : Mi(xly ' ' 'vxlyxi+1r ' ' yxn)' (1:= 1! ' . " n) (130)
What conditions are necessary on these functions or on.the func
tions f so that there exists a 2 as dened by equations (127)
and (128)? ,
Dene
" For example. U) = 0. (i = 1, ' ' '. 7!)or FU) = 0, where F(0) : O; F(a) # 0'
a # 0.
54 I'OUNDATIONS OF ECONOMIC ANALYSIS
The X's are unambiguously determinable independently of the

representation of the f s. In general,
kt} # X,
Ii there exists a function z. the vanishing of whose partial deriva

tives is equivalent to the equations (125), then
ku = % (132)
Also for every possible triplet
). l\ )i . = f ufcf 1:
= " fk f k"
, t * fnofuofkko fuufk'co n = Ankh-[Am (133)
because f., = f,. for all pOSSiblepairs

These conditions are necessary They are not identities for
all values of (xl. , x),but only for those satisfying
f(x;, . ,x) = 0, (z = l, , n)
i.e., they hold only at the point (x10, . x,.)" They are not to
be confused with the integrability conditions of chapter v Though
necessary, they are probably not sufcient Quite possxblysimilar
point reciprocity relations can be found between various derixatives
of higher order If all of these should hold, perhaps a complete Set
of sufcient conditions would be attained
The relationships given in equation (133)are vacuousif the num
ber of variables is less than three All together n(n 1)(n 2)/6
independent conditions are involved
" In addition, for all values of (x) satisfying a subset of r(< :1)equations f' = 0.
there will be dened similar relations which are identities in the remaining (n - r)
variables
Any completely non-speCialized set of 2:equations in 71variables. f (xi. . x..) = 0,
can be regarded as equwalent to a stationary posuion of a function of Zn variables
Let
!..(xl! , x'll xl HI exit-Im): z f.(vlo . &)an

!
dF = 0 implies among other things that f(x.. , x,.) = O This is denitely not an
extremism pasition as shown by reference to the secondary conditions The fact that
in a larger set of variables a stationary value corresponds to a non specmlizedsystem
seems devmd of economic Signicance See G D BithoH, Dynamical Systems (New
York 1927). Pp 3334
As we have seen, it is not enough that our equilibrium equations
be expressible as the partial derivatives of some function. In order
to be able to derive denite theorems it is desirable that this func
tion be at a regular maximum or minimum. This requires as is
shown in Mathematical Appendix A, Section II, that certain
quadratic forms be either negative or positive denite. Thus, for
a maximum
>":fa-hihi < 0, not all h.- = o. (134)

1 1
As shown in Appendix A, Section III, this requires that the follow

ing inequalities hold.
. fn fn . fn fn f13 .
lf11l< 0. f f > 0, fai f22 fza < 0, etc. (135)
f fa: fs:
For a regular minimum these determinants are all positive. It can
be easily shown that either of these is equivalent to the following
conditions.
1 " 1 Mi MI:
I) 1 >03 "ii 1 )jk >0; etc. (136)
" )- xk, 1
for 13,j, and I: all unequal. This proves that it is not important
that our problem be either a maximum or minimum one, but only
that it be one or the other.
As an example of a problem which can articially be converted

into an equivalent maximum problem, consider a number of inde
pendent rms buying the same kinds of productive services in
purely competitive markets.18 The demand of any rm for the
factors of production can be written
v.- = 7"(191. Pn). (i = 1. , n) (137)
where (vl, . . -, v,.)represent respective amounts of 1;factors of pro

duction, and (p1, - - -, p,.) their respective prices. It can be shown
" H. Hotelling, Edgeworth's Taxation Paradox and the Nature of Demand and
Supply Functions," Journal of Political Economy, XL (1932), 577616. L. Court, "In
variable ClassicalStability of Entrepreneurial Demand and Supply Functions," Quarterly
Journal of Economics, LVI (1941), 134144. R. Roy, De lUtilil, Contribution 41la TMarie
des Choix (Paris: 1942).
that
19-.: &.
ap; " ap; 383
Where we sum over the individuals and use capital letters for the
total amounts demanded
K=Zv.=R(P:.".1-). (i=1.---.n) .(139)
From (138)
aR R'
ap, " ap. )
There exists, therefore. a function
Z = RG1! ' : pu) .- (V1P1+ Vng + ' + VRP-). (141)
where
3613
?. = Rm. m. (.- = 1. n) (I42)
The conditions of equilibrium represented by the general demand
functions of (139) are, therefore, equivalent to those derived {tom
the condition that Z be at a maximum with respect to the 9s, 1.e.,
Q.=R._v,ao. (a=1__,,) (143)

ap;
CHAPTER IV
A COMPREHENSIVE RESTATEMENT OF THE
THEORY OF COST AND PRODUCTION
ECONOMIC
THEORYas taught in the textbooks has often tended to
become segmentalized into loosely integrated compartments, such
as production, value, and distribution. There are, no doubt, peda
gogical advantages in such a treatment, and yet something of the
essential unity and interdependence of economic forces is lost in so
doing. A case in point is the conventional assuming of a cost
curve for each rm and the working out of its optimum output with
respect to its demand conditions. Only later is the problem of the
purchase of factors of production by the rm investigated, and
often its connection with the previous process is not clearly
brought out.
I should like here to investigate from the point of view of the
previous chapters the cost curves of the rm as usually presented
and the production function embodying technical relations between
inputs and output which lie behind it, and also to show clearly its
relevance to the problem of the determination of the optimum
output. In particular, I shall attempt to derive all possibleopera
tionally meaningful theorems. Much of what is said will be found
to apply regardless of the elasticity of the demand curve for the
rm; i.e., under impure as well as pure" competition. By em
ploying a suitable notation it is possible from a purely technical
point of view to consider the case of any number of productive
factors as easily as one or two.
STATEMENT OF PROBLEMS
In the beginning the revenue side of the rm is completely
ignored. We assume as given by technical considerations the
maximum amount of output, 3:,which can be produced from any
given set of inputs (vl, - . -, vn). This catalogue of possibilities is
the production function and may be written
x = (v1, - - v, Un).
57
58 FOUNDATIONS OF ECONOMIC ANALISIS
In general there WI" be a maXImum output for each set of Inputs

and so thus function Is smgle valued and null be assumed InItIally
to hate contmuous partIal derIvatu es of deSIredorder Further
more no marginal phySIcal productmtIes can be negatne else
output uould not be mammal smce It could be Improved for the
same set of factors by leavmg some Idle Regardless of the, con
sxderatIon In the relevant regIon under conSIderatIon ue wullhave
@. 0 (t = 1 n) (2)
where as a conventIon of notatIon
83:
99.= 5; , = margmal phySIcal (degree of) productIVIty SImI
larly, ue dene
_ 61:
. " v av,
It 15also assumed that each rm 15small relatIve to the market
for each Input so that IInhmIted amounts of each can be bought
at the respectIve prices (w, w)
As a matter of denItIon the total cost of the rm may be
ertten as the sum of the costs for each Input Item and any other
costs thch are Independent of the purchase of the deSIgnated
Inputs and output I e
C = A + ww. (3)
where A represents costs whnch do not vary WIth the deSIgnated

Inputs and output (taxes etc) Of course these xed charges
may be zero
A brIef survey of our eld by economIc IntuItIon wrll aId In
phrasmg the problems to be nmestIgatcd after \IhIch mathematical
analySIs may be employed In statmg the condItIons Imposedon our
varIous functions Our mm 15to denve the total cost {or each
output More precrsely, thh gIven pnces of productIve factors
and gwen productIon functIon we are Interested In denvmg the
mzmmum total cost for each output Tlns I" be a functIon as
follows
(, = A + V(x w. w..) (4)
. [[ e regard the puces of the productIve factors as constant the

THEORY OF COST AND PRODUCTION 59
resulting relationship between C and x is the usual total cost curve,
from which average and marginal curves can be derived.
The question arises as to the relationship between (3) and (4).
It is clear that the same output can be produced by an innity of
combinations of productive factors so that in the absence of other
considerations it is impossible to determine the total cost uniquely
for each output. In view of the consideration that total cost for
each output be a minimum, our indeterminacy disappears. Out
of the set of all possible input combinations which will produce a
given output, that combination is selected which minimizes the
total cost dened in (3). In other words, with given prices of pro
ductive factors and for a preassigned output, there is an optimum
value for each productive factor; i.e.,
v,-= f(x,w1,--,w,.). (i =1, ---,n) (5)
By substitution into (3), we have
C = A + iw;f(x,w1,---,w,.) = A + V(x,w1, '-,w,.). (6)

1
Thus, the relationship between (3) and (4) is revealed.

It is part of the purpose of this book to investigate the proper
ties of the functions (4) and (5). It is true that theoretical eco
nomics does not deal with particular forms of functions (e.g.,
polynomials, etc.). However, it does concern itself with the
general character of various functions; that is, their slopes, curva
ture, etc. In this case we are interested in the followingproperties
of these functions:
ac a2c ac 62C _ (7)
_x 513' Eu? 6x610,"
i.e., how are total and marginal costs affected by changes in output
or prices of productive services? On what properties of the pro
duction function does this depend?
We are also interested in
i ; (8)
67.0,- 6w,- ' x
i.e., what is the reaction of demand for a productive service with

a change in its own price? With a change in another price? With
a change in output?
These are obviously important theoretical questions. and yet,

curiously enough, the answers to some of them do not appear to be
in the literature. In the next section these will be mathematically
evaluated, and the results summarized. It will be found that
under very general assumptions the knowledge of the signsof these
differential quotients will give us the direction of change with
respect not only to suiciently small nite movements, but also to
nite movements of any size.
CONDITIONS or EQUILIBRIUM
Thus far we have employed only the notation of mathematics.
By so doing, the problem has been clearly framed, which in many
economic problems is more than half the battle. It remains now
to state the derivations of (4) and (5) from (1) and (3).
Our problem is to minimize
C=A+iww. (9)
l
subject to
@(01, ' , v..) = a:= constant. (10)
Mathematically, this is a constrained minimum problem, and we
may avail ourselves of the method of the Lagrangean (undeter
mined) multiplier. We dene a new function
G = A + a, A[p(v1, --, v..) a]. (11)

where ( ).) is a Lagrangean multiplier, whose economic interpre
tation will be brought out later. G may be regarded as a function
of all the inputs as independent variables, It is necessary for a
proper relative minimum that
85%=%='w3V1.. {\;_1l,...'vrl (12):
These may be rewritten as

l===...=. (13)
101 w: wa
This is the well-known economic theorem that in order for totoltosts

to be a minimum for any given output, the marginal productrmty of
the last dollar (1A) must be equal in everyuse. Alternatwely, It may
be stated that the marginal physical productivity of any factor
must be proportional to the price at which it can be hired, the
factor of proportionality being the term A} It will be noted that
this condition is independent of the revenue curve of therm and must
hold at every point on the cost curve, not only at the nal point of
optimal output.
Provided that certain secondary conditions to be discussed
presently hold, the n equations in (12) and the one equation in
(10) are sucient to determine each of our (11+ 1) unknowns
(vl, ~-, 11>.) in terms of x and the factor prices (wl, - - -, w") re
garded as parameters. Thus, our equations (5) are dened im
plicitly by these minimum conditions.
One may well question the advantage of this formulation. Ap
parently we have substituted an indirect statement of our nal
conditions for a direct one. But, and this is typical of.all sound
economic theory, we are trying to deduce the consequences of our
hypothesized data, and we do know (by hypothesis) much about
the functions in (12). Merely,to state the relations in (4) and (5)
is formal and empty. By means of (12) we can place positive
restrictions upon them and know their general properties.
Now there remains the purely mathematical problem of trans
lating our assumptions into terms of the functions (4) and (5), and
computing their respective partial derivatives designated in (7)
and (8).
SECONDARY EXTREMUM CONDITIONS
To do so we must rst state all that we know about the condi
tions dened in (12), in particular the secondary necessary and
sufcient conditions for a proper relative constrained minimum.
It is clear that in order for total cost to be a minimum for a pre
assigned at,the locus of all possible inputs yielding that preassigned
quantity (isoquant surface) must be tangential to a locus of all
possible inputs yielding the same total cost (isocost plane). But
of course this is not enough. The isoquants must also be convex
to the origin in all directions in order that its contact with the
isocost plane represent a true proper minimum. The analogy with
the theory of consumers preference suggests itself. This is brought
out more clearly if we phrase the problem not as that of minimizing
' Later >.will be shown to be equal to marginal cost, i.e., BC/ax.
(.
total cost for a preassigned output, but rather in the equivalent

form of maxxmizmg output for any preassigned total expenditure
Mathematically, our secondary conditions are
Zx Z1 (0115351< 0! (14)
for X
.. L/
? PIE: = ou
and not all
! : 0
Con51derthe bordered determinant
9011 Pu (Din 991
35" 52. 21 9922 Wu 90:
= (O.; P. =
D l P: 0 (15)
051 ion? Pn Pn
so: 992 p.. 0
and respective principal mmors

@ 11 (p 12 Pt Pu (P12 (.913 901
D: Pu 022 02 . 012:: le 9022 6923 02 . etc (16)
9931 $932 (033 99:
sal <02 0
m sa: sa: 0
It is well known that (14) implies that any such minor of the
order m must have the sign of ( l)'l, and conversely. ie ,
(- DMD .) > 0 (m g n + 1) (17)
Specically,
Qu?) "'" 2PUSI'P;+ 97139032
< 0; etc (7' 76.1.)
It ill be noted that this condition does not necessarily imply or
require the law of diminishing marginal physical productivity
to hold "
' Compare H Hotellmg "Demand Funcnons WithLimited Budgets." Etonvmetnca.
January, I935. pp 66-78
i am assuming that these secondary conditions hold not only at the minimum
_
Point. but everywhere Mathematical!) . this assures us of the uniquenessof our equl
: -. smce this stronger _ V., :_" _ :, les out multiple relative mmuna
The secondary conditions are not always mentioned in the
literature. It is not for elegance or completeness that they are
included here, but rather because they are wholly relevant to the
problem under discussion, since it is upon them that all our results
depend.
DISPLACEMENT OF EQUILIBRIUM
It is now possible to derive in compact form the rates of change
of our dependent variables (vl, - - -, vn) with respect to changes in
the variables (x, 101,-- -, w"). The reader not interested in the
mathematical derivation of these conditions is referred to a sum
mary of results at the end of this section. First we write the total
differential of our equilibrium equations (12) and (10).
div,
z":.-,dv,-+%'dx=T.
1
(i=1.---,n) (18>
, gaJdv, = dx.
1
These are (72+1) linear equations in (11+1) unknowns

(dm, - - -, dvdx), and may be solved in determinantal notation as
follows:
n d i
2:71" A+ dxAn+1.L
d'il}; = l A , (19)
where

(Pu 9912 99m
Pz
ga-l
I, 99- _- (pal
o 9022
. (.021:
. )
.
A : gai _- _ _ . . (20)
O .
(Pn! 90112 Pnn
901 4.92 ' 99n 0
and A.Iris the cofactor of the element in the rth column and the
qth row.
Also,
" dw,
dh = zx:T A +1+ dard, n+1
(21)
Hence,
'._ &
w, _ XA (22)
As a specnal case
59.: _ ka
awk ... A (23)
Also,
% Ami1k
69: A (24)
and
_a_A__ A} n+1
awk _ RA ' (25)
Q = Amt-ln+l
x A (26)
It 15clear on mspectxon of the determmant A, that

1
A KD (27)
Also
1
A =XD:k=Akf (J.k=1. .") (28)
Ai n+1 = D: n+1 = XA: (J = 1. s) (29)

and
Ami-ln+1 : Dn+l n+1 (30)
Therefore, from (28)
935-2,. (31)
day,awk
That 15to say, the change m the kth factor wzth respect to a change
m the 3th pace, output bezng constant, must be equal to the change m
the Jth factor wzth respect to the kth pnce, output bemg constant, a
result whxchxsnot mtumvely obvmus
From (27) and (28)
(32)
But from our stability conditions in (17)
D:" .
<0' (J=11"'rn) (33)
Therefore,
av,- .
rw,- < O. (J = 1, -, n) (34)
That is, any xed output will always be produced with less of any
given factor as its respective price rises, other prices not changing.
By the law of the mean this can be shown to hold for nite changes.
Let us now determine the economic meaning of A. Rewrit
ing (11)
G = A + fwd); _ ESOJI,
"'!vn) ' 5]:
1
and differentiating G, which is total cost with a term added on,

we get
66
8-3; .
This suggests that X may be marginal cost. This can be proved

rigorously in two ways. Of course,
dC = w,dv,-, (35)
1
and
dx = i tom. (36)
].
Dividing (35) by (36), we get

w,dv,
: :_ . (37)
ax Z Ptdvi
!
Substituting from (12),
Mada),- .
C __l____ = Aer, w!) . . ., w"). (
ax _ tpzdvi
l
66 FOUNDATIONS OF ECOAOMIC ANALISIS
More rigorously the proof xsas follows
EE .. 93
6.1. _ . to, 6x (39)
Substituting from (24),
_ Art-F1I _ * An+1 .
BJC21:10. A >\Zlcp. A (40)
But expanding A m terms of the elements of the last row,
A = &? @lAn+l I (41)
Hence, C A
& = XE = X (42)
Therefore, we may rewnte (12) as follows
6C
20. = a Pt (43)
Thus, xt may be stated as a theorem that m orderfor total coststo

be a mmzmum for any gwen output, the przce of eachfactor must be
equal to margmal physzcal productmty tmzes margmal cast Tlns
holds regardless 0j revenue conszdemnons
Of course,
8% C
& = 3f-r (44)
and
2
m; : exam
From (24), (25), and (29),
vk
_wg" ax (46)
or
2
C 02": (47)
exam= 5
Thxs has been pomted out In another connecnon m lectures by Professor Vmer
wnh rllummatmg mslght mto the relatronsth of the external to the Internal margm and
the broad zone of Indlflerenceas a subsntnte l'or the mmteslmal Pendant-ally, It u
du: condition whtch ISbeen: to Mr Wong : famous envelope theorem]
That is, the change in any input item with respect to an increase
in output must be equal to the change in marginal cost with respect
to a change in the price of that input item.
Recall from (26)
ax _ 62C A
ax_ a? T ' (48)
Now it is known from (17) and (27) that A has the sign of ( 1)".
Also
sen sa
Am. = - - = H, (49)
sent sa.
where H is called the Hessian determinant of the production func

tion. Obviously, therefore, the slope of the marginal cost curve must
have the same or opposite sign as compared to this Hessian, depending
upon whether the number of inputs is even or odd; i.e.,
62C
( 0an?? > 0. (50)
Thus the stability of pure competition is intimately tied up with
the Hessian of the production function, a result not intuitively
obvious.
I should also like to indicate certain other results, leaving their
rigorous derivation to the interested reader. Consider equation (5),
v=fi(x1wlv''rwn)' (i=1,--,n)
These functions are dened by (10) and (12),
x = 99(in ' '! Un)
and
wiXga,=0. (i=1,-~-,n)
(12) may be rewritten as follows:
$391 - (i =21'1n) (51)
01 101
Obviously, the changing of all prices in the same proportion will

leave the solution of (51) unchanged; hence, the function
must be homogeneous of order zero in the variables (w., - - ., w..)

x being constant; 1e .,
==f'(x.w1.-,'w..)=f'(x.7w1."'.7wn). (i=1. -~-,n) (52)
where yis arbitrary. Hence, from Euler' s theorem on homogene
ous functions.
: ;}...3; (a: 1, --.,n) (53)

This can be veried by substitution (rem (22).
By analogous reasoning,
" C
C_A+z.w'w: (54)
where
_C
6211;
=v.. (55)
Actually, from the considerations of the previous chapter it is

possible to deduce a more general condition which includes (34)as
one minor part. The minimization of total expenditure for a given
output, price being xed, implies (10) and (12). In addition, at a
regular minimum we must have
" av.
EI 25-
was, < 0. (56)
for not all . proportional to w.. Consider the determinant
6(2)]! ' v vn) at"
G 6(w1 . . . w) o w, . ( )
and
a_ .621
am....) o_w.aw. _ _ .. 15.3.
Gn 6(101, 2112): 50: l! , GI _ (wh 102,203)' Q . \ \
620] aw:
Then each such principal minor of order m (less than 7:)must bc

negative or positive depending upon whether m is odd or even; i.e..,
(- l)Gu...... > 0. Gun... = G = 0. (59)
Campan: chap 111,p 30, Mathematics! Appendix A, Sections IV and V, and chap V.
pp 113-116.
Specically,
_av,- < . _!
v __
av], _ ___?
69- 2 .
a'wj , wj awk (awk) > 0' etc. (60)
It may be well to summarize our results of this section:
621- a i!
4<
wf 0; M)
6(w1') wk) >0; etc. (34)and(59)
ave _ %
aw,- _ wk (31)
62 _ al,
xw; _ x ' (47)
6C .
w.- = Mp.- = 3; a.-, (z = 1, -- -, n) (12) and (42)
FC
( 1)H-x2< 0, (50)
" v; .
jwj=0, (z=1,--o,n) (53)
C
&: = I" (55)
" 6C
C=A+lewrawi' (54)
BOUNDARY OR CORNER MINIMA
Even in the case where the production function and its deriva
tives are continuous with the proper convexity to insure a uniquely
determined optimal position, the interesting case may arise where
some factor may not be used at all. That is to say, the more that
others are used and the less of it, the lower will costs be for any
given output. In this case the conditions for equilibrium do not
require the equalization of the marginal productivity of the last
dollar spent on it to that of other factors. Rather do we have a
boundary minimum due to the fact that no negative values are
economically admissible. Hence, the conditions of equilibrium
are given by the statement that for any input, potentially usable
but not actually used, the marginal productivity of the last dollar
7O FOUNDATIONS OF ECONOMIC ANALYSIS
spent on it must not be greater than the marginal productivityof

the last dollar spent on factors which are used
Mathematically,
m 1
la..% X , (61)
where the uth factor is one not actually used.

As the price tu..changes, it may still remain unused until some
critical level at which it will begin to be used, and hence Will{all
into the analysis of the previous section. Of course, the critical
level may very well depend on the scale of operations; i e , output,
so that with the same price it may still be brought into use by an
increase in output "
The demand function for such a factor of production will have
the following properties.
: f(x. wwt, , w"), (62)
23;s 0 in some domain dened by Mac,ww . . ,wn) < 0 (63)

av"
8? < 0 in some domain dened by Mac,ww, , w.) > 0 (64)
where & is so constructed as to form the locus of all critical points
described above.
DISCONTINUITIES IN THE PRODUCTION FUNCTION
I should now like to drop the assumption that the production

function is necessarily continuous with continuous partial deriva
tives at every point This assumption has been challenged by
many economists, who have alleged that production coefCients
' There is an interesting discussion of an exactly analogous eqmllbum SYStem!:
the famous paper of I Willard Gibbs, The Equilibrium of HeterogeneousSubstances
CollectedPapers, I, 55-349 '
7Anuntil
budget analogy is suggested
income increases to
or the caserelative
their of Itemsprices
which do not enter
decrease intocritical
to some a consumeil'
leve $5
Be it noted that this phenomenon here described can occur even though there be '
creasmg marginal physxcal productiwty, Just as the budget case does not relstrict :ihe
behawor of marginal utility It can be mathematically proved that this resu t 13:3:
pendent of the cardinal measure of product (utility) AnOther analogy 15971" Y
the classical comparative cost doctrineaccording to which a country 5913 izes C;
pletely m one good, the equilibrium is dened by a certain inequallty between Pm
and marginal costs
THEORY OF COST AND PRODUCTION 7I
are technically xed, that some factors are limitational, some
factors of production must be used in certain joint proportions,
etc. These discontinuities, if true of the real world, have been
thought by many economists to offer serious problems to the analy
sis of distribution and the pricing of the factors of production.
It will be argued here that the fact of discontinuity offers no
problems to the rmon the contrary, its task is made much
easier. As an obiter dictum I hold that it also o'ers no particular
difculty to the analysis of the wider problem of determining the
prices of the factors of production with which each rm is to be
confronted. As before, these are to be determined by general
equilibrium analysis of supply and demand.3
As before, we have a production function relating maximum
output to any given set of inputs:
x = so(v1, vn). (65)
Precisely as argued in the rst section this must be single-valued.

However, it need not be continuous nor have partial derivatives
at every point. In order that there be no contradiction to our
denition of output as maximal, we must have
Ago ?: O, for Av.- g 0. (z' = 1, ---, n) (66)
That is, as we increase all factors together, output cannot decrease,

since otherwise product would not in the next position be maximal.
For deniteness I assume that along an isoquant the PTOdUCtion
function contains only a nite number of POntSWhiChdo ?
possesscontinuous partial derivatives. At a point of discontinuity
it is assumed that left-handed and right-handed partial derivatives
exist. Of course, at a point of discontinuity there does not ?}ISt
a uniquely dened plane tangent to the isoquant, bUt hmltmg
direction cosines can be found for all planes WhChtough but do
not cut the isoquant surface. It is also assumed that the isocll-aftS
are of a single concavity," to be dened later. The production
function so dened is general enough to include the case Of xed
coefcients of production, perfectly complementary " factors,
limitational factors, etc.
' It is possible that within a narrow range the price may be indeter'n _
Specialses to coincidental inelasticity of supply and demand.
Note that all the inputs are to be regarded as independent

variables. It is never true that they must be used in xedprepor
tions. It is true that it may be unprotable not to do so, but this
is a result of economic calculation. Even in the continuous case,
given certain economic data, the factors must (on considerationof
protability) be used nally in given proportions, and in deter
mined amounts for each output. The only difference between
these cases is that in the discontinuous case the required optimum
point may be more obvious and less sensitive to changes in the
prices of all factors of production. Be it understood that I do not
minimize discontinuity. On the contrary, I should like to indicate
the outline of a method which will handle both cases.
Along an isoquant it is assumed that we have convexity as de
ned below. Consider any point on an isoquant, (08, -.v,.).
There must exist constants, (m', - o.cal), (not necessarilyunique)
such that
use.2 9.1) a o, (67)
I
where (vf, .. -, 12.2)is any other point along the same isoquant.
This merely says that there exist one or more tangent Planesat
each point which touch but never cross the isoquant. Similarly,
at the second point there exist constants, (af, . -, au:). sud! that
f(v. v.) 2;-o, (68)
01' ;
aew v.) a o. (69)

Adding (69) and (67) and changing sign, we get
in: www go. (.10).
1
Since these are two arbitrary points, taking one as xed, say
(vf, - - ,Uno),we must have along an isoquant
a,Au. :$ O (71)
l
{or
Olo! . ' ': vo) : 500,10+ Av! ' ' ': au + Av"):
and
>:
1
AAU. 0. (72)
Actually, it can be veried that the necessary and sufcient
l
conditions imposed on (a1, ---, an) in order that (71) holds are
the following inequalities:
(Pis S Oli SoiL
IM , (73)
SOIL 010 %s
where go.-"and go.3

are respectively the largest and smallest of the
left- or right-hand derivatives at the given point. At any point
where there is no discontinuity, the right-hand and the left-hand
derivatives are identical; hence, the inequality converges to the
equality,
Clio i(910, ' ' ' , Uno)
Of1_ %(lo, ' ' ', no). (74)
All of the above are merely elaborations of the denition of con
cavity. It remains to show its relevance to the problem at hand.
CONDITIONS OF EQUILIBRIUM
Suppose we are given a set of prices (101,' ' ', 'wn")correspond
ing to which there is one combination of factors (11?,- - -, v,.)which
minimizes total cost for given output. As a denition of our
minimum
AC 0 (75)
for
Ax = 0, Av.-g 0.
In other words, for any point (vl, --, v,.) along the same isoquant
we must have
w.v _>__ u.-? (76)

1 1
for
(vlg . . ., vn) : P(v10, . . ., vnO)
or
tv,-"Av; 0
l
74 FOUNDATIONS or ECONOMIC ANALYSIS
for
4910+ Avg, - . -. u." + Av..) == (vi. ---. vn)
Obviously from (73) this implies
geg<ei
8
'PIL _ w1 : Ii3 (78)
In the case that the maximum point is one at which continuous

derivatives exist, we have the conditions of the previous section,
&_m
P: h 101" (79)
Thus, condition (78) is the general condition which includes (79)

as a special case.
Furthermore, it is still possible to place denite restrictionson
the demand functions for the {actors of production, output being
constant. As before,
1 : f.(xl wlov ' ': who)! (80)
or
Av. = Jlx. w i-2m. -. "'- 4- Aw.) - v.. (81)
From (72) we have
mm. 2 0 (82)
1
for
Ax = 0.
Suppose that only the price of the lath factor varies. Then
(82) becomes
4103A; 0. (83)
That is, an increase in the price of a factor cannot result in an
increase in its use. Likewise, a decrease in its price cannot result
in a decrease in its use. Still more generally, it can be stated that
a change in the price of any number of factors cannot result in a
change in amounts of all the factors in the same direction. i.e.,
Aw,Av, ; 0, (r =_5n) (84)

l
{or all Aw, 0, not all Av, > 0; likewise. all Aw, g 0 implies not
all AU}< O.
DETERMINATENESS OF EQUILIBRIUM
Clearly our minimum cost is unambiguously determined even
in the case of discontinuity. The task of the rm is easier because
the penalty for not being at the minimum point is greater and more
obvious. To be at the top of a smooth hill requires ne balancing
and judgment. To nd a maximum which is a cusp or spire is
much easier. Moreover, such an equilibrium is extremely stable.
It is sometimes called too stable equilibrium. In order to move
it a very large change in prices may be necessary. In the limiting
case where the increase of each factor in more than the optimum
proportion results in no increase in product, the wage may uctuate
from zero to innity without changing the proportions of factors
employed for each output. And yet the problem is completely
determined.
It is curious to see the logical confusion into which many
economists have fallen. The primary end of economic analysis is
to explain a position of minimum (or maximum) where it does not
pay to make a nite movement in any direction. Now in the case
that all functions are continuous, it is possible as a means towards
this end to state certain equalities on differential coeicients which
will (together with appropriate secondary conditions) insure that
certain inequalities will hold for nite movements. It is no exagger
ation to say that innitesimal analysis was developed with just
such nite applications in view. Unfortunately, the means have
become confused with the ends, and so conventions and artices
are continually sought in order to be able to make statements con
cerning marginal equivalences. A case in point is the Marshallian
marginal net productivity. It is only in the singular case that the
Production function is differentiable (i.e., possesses certain prOPf'
ties of continuity) in a certain direction (i.e., for certain compOSIte
movements) that it is possible to employ this device; whereasothe
inequalities of condition (78) always give necessary and 511339th
Conditions, and include the marginal net productivity relations
whenever these happen to be applicable.9
' An example of another improper use of marginal net productivity curves ISpro
VidEd by the treatment in Mrs. Robinson's The Economics of Imperfect ICompetttzon
(London: Macmillan, 1933), chap. xx. The reasoning is mathematically Circular; the
50-lled demand curves shift with changes in the wage of one factor! This followsfrom
the faCt that they are drawn up as of appropriate amounts of the otherfactors. These
appropriate amounts are necessarily a function of the prices of all factors of produ-
Although equnhbnum ns determmate for the smgle rm con
fronted wrth puces of the factors of productlon dlscontmumesof
the productlon function may Introduce some dlicultres mto the
general equnlrbrlum problem whereby all rms and 1nd1v1duals
together through the Interplay of then demands and supphesde
termme the prrces thh whlch each ns confronted For drscon
tlnurtles may mtroduce perfect melastncntresof demand ln certam
domarns, there emerges the possrbrlrty, however remote, of come:
dental melastzcntresleadmg to mdetermmacy of price wrthm certam
lxmnted ranges But thus IS out51de the sc0pe of the present
dtscussmn
MAXIMIZATION OF PROFIT
By now we have attamed to the pomt where most discusstons
begnn We have seen how to derive the locus of factor cornbma
trons whlch grve lowest total costs for each output But as yet
the scale of operations level of output, actually to be undertaken
has not yet been determmed This can only be done m the face
of a new set of consnderatnons, those relatmg to the terms at whrch
different amounts of the commodity can be sold I take as a
datum the maxrmum amount of gross total revenue WhIChcan be
secured for each level of output Thus may be wntten
R = R(x) (85)
Let us dene prot, net revenue, as the difference between gross
revenue and total expenditure,
7 = (xr w! | wa) : R(x) _ A "V(x 101, I w") (86)
Output wl be optlmum when prot 15at a maxrmum Necessary
condltlons that thrs be so when all functlons are dl'erenttableare
%; = 52%
g o (88)
Assummg that we have a regular relatrve maxnmum, thus becomes

85 = (89)
x x '
I
ax2(?a", (90)
This is the familiar theorem that at the optimal output the
marginal revenue curve must intersect the marginal cost curve
from above.10
Regardlessof whether or not the prot function has derivatives
at each point our proper maximum conditions are
A1r < 0, for Ax Z O. (91)
or
AR AC
E < E , for Ax> 0, (92)
AR AC
Z; > ; , for Ax < 0. (93)
The economic common sense of this is obvious.11
Equation (89) gives us a relation to determine optimum output
x. By substituting the value of x, so obtained, in (5) we get a
new set of demand curves for factors of production, drawn up as
of a given total revenue curve.
v.-= f(x,w1, - - -, w") = g('w1, -- -,w,,). (i = 1, ", n) (94)
It is possible to derive the positions of output, input, etc., in a
more direct fashion by treating all the inputs as independent
variables.
Let
(1):,.. ~,v,,,'w1,. . .,wn) = REM, .. 371,0] _ A _ wi' (95)

At a proper maximum
gigging-wee. <=L~~n> <96)

and
T = [Ruby + szioiof]
must form the coefcients of a negative denite quadratic form.
1The famous cost controversy may be interpreted as an argument over the impli
cations of these conditions.
these condi
u In the case of pure competition, when price is independent of sales.
tions become
AC
< and > AC
- , . ly.
respective
p A:: p Ax . 'b
It can also be shown that Apr g 0; i.e., that an increase in prices cannot. cete!'5 Pant ;;
result in a decrease in quantity supplied. Hence. the fm's supply curve canno
negatively inclined.
The results m (96) follow from (43) and (89) smce for any
output factor price must equal margmal cost tlmes margmalphys
real productmty whrle for Optimum output marginal cost equals
marginal revenue These are n equations from whrch we can
solve for the n factors of production 1nterms of the n puces to get
the demand functions of (94)
I). ::= g'(w1 . un) (Z= I, , )
Actually, 1t lS known that "
_6v. I2
6w, = gli : T : (97)
where T., 15the cofactor of the element of the zth row and the 3th
column of the above mama From the denntenessof the above
quadratrc form It follows that these last must form the coefcnts
of a negatrve demte form, 1e ,
g; 6(g. g")
aw, < 0, (TH(w,20;) > 0, etc (98)
INDETERMINACY IN PUREST COMPETITION?
If competrtxon IS pure" m the commodlty and factor markets

and the productlon functlon rs homogeneous of the rst order then
It 18a classrcal fact that the matnx T, Inch except for factors of
proportronalrty 15by the rst assumptron 1dent1calw1ththe Hessran
of the productron functron 15sxngular Therefore, a regular max:
mum for the rm IS 1mposs:ble Umt costs being constant and
demand bemg horizontal, there are only three pOSSlblfltleSpnce
bemg everywhere greater than marginal cost, 1t wrll pay the rm
to expand Indemtely xe , untrl competrtron ceases to be pure or
1fprlce rs less than marginal cost no output wrll be produced or.
nally, 1f price rs Identrcally equal to margrnal cost the exact
output of the rm wrll be a matter of 1nd1fference Thus what
was usually regarded as the most favorable case for pure compet!
tron turns out to yield mdetermmate output for each rm
However, too much should not be made of thusparadox Even
though the output of every rm may be mdeterrnmate therr sum
" Compare H Hotell ng Edgeworths Taxanon Paradox and the Nature Of
Demand and Supply Functrons Journal of Polmcal Economy XL 577-616
may be determinate, in the same way that the sum of two discon
tinuous functions may be continuous. For if many rms expand
output, others will contract; for price will fall along the industrys
demand curve causing contraction. It has been argued, however,
that competition will disappear since any one rm, with no ob
stacles to its expansion, will grow until it is a signicant part of
the market, at which time it will be able to a'ect the price of
its product.
This is similar to the familiar argument by which it is shown
that decreasing marginal cost within a rm will lead to monopoly.
Nevertheless, the analogy is faulty. For the demand curve of any
rm is equal to the demand curve of the industry minus the supply
curve of the remaining rms, already in the industry or potentially
therein. This being the case, it is easy to show that under uniform
constant costs the demand curve for a rm is horizontal even
though it produces 99.9 per cent of all that is sold, Geometrically,
the long-run supply curve of potential rivals is horizontal, and a
horizontal curve subtracted laterally from any curve must always
yield a horizontal curve. Economically if the rm were to begin
to restrict output so as to gain monopoly prot, it would cease to
sell 99.9 per cent of the output or even anything at all. Conse
quently, it would not attempt to do so, but would nd its maximum
advantage in behaving like a pure competitor.
Thus, it remains true that the classical assumptions underlying
pure competition are actually consistent. It is no accident that
Walras and Marshall paid so little attention to the rm and so
much to the industry. For under the purest conditions of competi
tion the boundaries of the former become vague and ill-dened, and
also unimportant, since through reactions to prices the factors of
production adjust themselves in the right proportions and in the
right total amounts for the industry. .
Perhaps to a greater extent than in the case of increasmg.cost
will the industry be subject to oscillation around its equilibrium.
However, once displaced, output will tend to return to the correct
value, so that the equilibrium may be said to be stable even though
it would appear that a limitingly small individual would have no
incentive to keep his output unchanged- However, the same ls
true of the situation where the price of the same good m two
markets just balances the transportation costs necessaryto transfer
a unit of the good from one market to the other. Any one arbi
trager has no special inducement to ship more or less than he is
actually doing, nor even to ship just the amount that he is doing.
Yet the equilibrium is stable; for if the right amount were not to
How, the spatial price differential would change so as to return the
system to its previous position. To an innitely near-sightedolive
the bottom oi the cocktail glass appears level, and it no doubt
regards itself as being in neutral equilibrium. Actually, the equi
librium is stable as any nite movement will show.
DISCONTINUOUSCase
In the general case where the production function is not neces
sarily differentiable we have still for a maximum
Arr g 0, for Av. % 0. (99)
As a special case of this, for a movement of one factor. all others
constant. we must have
AR Ax ._
753:7. < w.. for Av. > 0. Av: 0. (10)
__
AR Ax = , l1
Av. > tv,, for Av. < 0. A: 0 ( 0 )
That is. it must pay to move neither backward nor forward.
Moreover, consider a given set of factor prices (201,- - o, 102).
Corresponding to this there is a set of factors (91,- ~-, v..)which
give a maximum prot. In order that this be a real maximum
RDPM. -. v.0] - A * );wfv. R[(vx,"'. la)]
"' A "' wtovuo- (102)

l.
Consider a set of prices (201,. - , m) _for which the (arbitrary)

point (m, - - , v.3) is the point of maxxmum prots. Then
REsowi". - -, v.)] - A - wvr" % Rpii. --.v..)]
_ A _ ww. (103)
l
Now adding (102) and (103) and canceling terms we get
$;: (w.1 wma,-1 ?).-) ; 0, (104)

or
ZZjAwiAv;g 0. (105)
For only the jth price varying, this becomes
Aug-Av;g 0. (106)
In other words, a decrease in a price cannot result in a decrease in
the factor used. Further possible interpretations are of course
possible.
As before, the general case is simpler than that of the special
continuous case. Moreover, the method of nite increments ap
pears to be mathematically simpler in the sense that it is possible
to state the qualitative direction of changes without solving in
versely for the actual demand functions.
The method employed here is that which underlies Le Chateliers
principle in physics. By making use of Professor E. B. Wilsons
suggestion that this is essentially a mathematical theorem ap
plicable to economics, it has been possible to gain increased gener
ality without increased complexity and emptiness.
It is important to realize just how much content there is to a
particular economic theory. As far as the single rm is concerned,
everything fundamental which can be said is implied in the state
ment that in equilibrium there must exist no movement by which
the rm can improve its prots; i.e., A7:- 0 for all movements Of
variables possible to the rm. In the case of continuity certain
necessary relations of differential coefcients (marginal equiva
lences) are implied. Moreover, assuming certain specic forms to
our functions (independence of prices, etc.), it is possible to deduce
formally the implications of an equilibrium position (e.g., negative
demand curves, positive supply curves, etc.). It appears that no
more than this can be validly stated.
EXTERNAL CONDITIONS or EQUILIBRIUM
Thus far we have been discussing the conditions of equilibrium
imposed from within the enterprise by its deSIre to maXImize
prots Thrs has resulted m certam margmal mequahtles Econ
omrsts have not stopped here, but have also tried to analyzecertam
condrtrons of equnllbnum resultmg from mtercompetrtxon among
rms That 15 to say, they have tried to state COHdIthHSupon
the market sxtuatxon (obstacles) thh WhiCheach rm M be con
fronted In particular, they have been mterested 1nthe determlm
tron of the rate of prot which any rm can earn
It has often been argued that not only must price (average
revenue) under perfect" competxtlon equal margmal cost. but also
It must be equal to average cost so that net revenue wnllbe zero
Thls second condrtron has not always been recognlzed as beingof
an entrrely different nature from the rst In tlus sectlon an
attempt wnll be made to dlStIHgUIShbetween them It IShoped
that m so domg xt Willbe possrble to put the famous addmg up"
problem and homogeneity of the productron functron m ntsproper
place
In the beginning, to avond confusmn no use 15made of the
term perfect ' competltron The term pure" competltronwrllbe
understood to mean that the demand curve for any producer 15
1nn1tely elastlc that his sales cannot affect prnces The problem
of drscontmurty 1s 1gnored Under these condttlons the Internal
condltxons of equlhbnum are that marginal cost be equal to mar
gmal revenue (pnce) and hence that the marginal physrcalproduc
trvrty of each factor tlmes the sales price of the good be equal to
the pnce of the respective factor
: : lg , (107)
ax x
C
w. = 5 P= Pv. (103)
These are margmal condltlons and say nothmg about the totals
'mvdrw. Lt.'sden. tat-m.h}; demrmn of. the long; run" as that
perrod m which all costs can be avorded by goxng out of busmcss,
that the rm must never have a negatwe net revenue As a
condmon mtemally Imposed we know that
fr 0, (109)
or
R(x) ; uw. (110)
l
Some writers, by a curious play on words, have been able to
arrive at the condition that average cost equals price. A typical
form of the argument is as follows: (1) a rm will equate marginal
cost to price; (2) it will also try to minimize its unit cost; (3) at the
point of minimum unit cost average cost equals marginal cost;
(4) hence, average cost must equal price (average revenue) and
prots will be zero.
Stated explicitly, it is obvious that the second statement is false.
The play on words arises from the confusion of the condition that
for each and every output total and unit costs must be a minimum
with the statement that of all outputs possible that one is chosen
at which unit cost is the lowest. The rst implies the valid condi
tion that the marginal productivity of the last dollar in every use
must be equal. The second implies the invalid (from internal con
siderations) condition that output be determined irrespectiveof the
sales price.
On the other hand, some have tried to argue as follows: (1) the
production function from the nature of things must be homogene
ous of the rst order; (2) by Euler's theorem it follows thatf if
factors are paid according to the marginal productivity principle,
product will be exhausted. .,
As an example of the lack of integration between the theory of
production and that of cost, we nd many writers asking whether
product will be exhausted at the same time that they have already
agreed that price equals average cost, and total revenue equals
total cost. Of course, the latter condition is merely another way
of stating the former.
Once the problem is properly stated as that of determining the
relation between gross revenue per unit and expenditure per unit,
it should be reasonably clear that this cannot be determined by
the properties of the production function alone, but must depend
upon the marketing situation of the rm, which in turn depends
upon the competition of other rms. It is quite clear that as far
as the single rm is concerned it is possible that it be making huge
prots regardless of the homogeneity of the production function.
This condition is neither necessary nor sufcient to the exhaustion
of the product. If the production function were homogeneous.bUt
demand were sufciently favorable, of course product wel
be exhaustedeven under pure competition.
The problem of homogeneity of the production function is one

about _whlchmuch controversy has raged. It has long been held
on philosophical grounds that product must be a homogeneous
function of the rst order of all the variables, and that if this is not
so, it must be either because of indivisibility or because not all
factors" have been taken into account. With regard to the rst
point, it is clear that labeling the absence of homogeneity as due
to indivisibility changes nothing and merely afrms by the implica
tion that indivisibility does exist, the absence of homogeneity.
With respect to the second point, we may reverse the Aristo
telian dictum and afrm that anything which must be true self
evidently (philosophically"), intuitivelyLe. by conventional
denition of the terms involvedthat such a principle can have
no empirical content. It is a scientically meaningless assertion
that doubling all factors must double product. This is so not
because we do not have the power to perform such an experiment;
such an objection is of course irrelevant. Rather the statement
is meaningless because it could never be refuted. in the sense that
no hypothetically conceivable experiment could ever controvert
the principle enunciated. This is so because if product did not
double, one could always conclude that some factor was scarce
It is useful, I believe, to avoid the expression factor of pro
duction" entirely. This has been used in at least two senses,
neither of which is quite satisfactory. First, it has been used to
denote broad composite quantities such as labor, land, and capi
tal." On the other hand, it has been used to denote any aspect
of the environment which has any inuence on production. I sug
gest that only inputs" be explicitly included in the production
function. and that this term bc conned to denote measurable
quantitative economic goods or services. The production function
must be associated with a particular institution (accounting,
decision-making unit, etc.), and muscuemwn'optrst'zmy'cnirqm
circumstances pertaining to this unit. Other denitions are of
Any function whatsoever in n variables may be regarded as a subset of a larger
functlon m more than 5 variables Wthh :s homogeneous of the rst order It ISbecause
thus :: true of any arbitrary function that this generalization ls useless For example,
the volume of a sphere not being a homogeneous function of the l'adllls,a new factor can
be dened whose "scarcnty" Willexplain this facr Lake a false proposmon logic from
which every proposmon an be derived, this overgenerality renders such a convention
useless.
course possible, but it is clear that our previous conditions cannot
be expressed in terms of them.
So dened, the production function need not be homogeneous
of the rst order. If really homogeneous, marginal costs would
always be constant. It is indicative of the lack of integration
mentioned above that many writers assume U-shaped cost curves
in the same breath with homogeneity of the production function.
In reality, it is not on philosophical grounds that economists
have wished to assume homogeneity, but rather because they were
afraid that, if they did not do so, contradictions would emerge to
vitiate the marginal productivity theory. This is simply a mis
conception as will be indicated below.
Our discussion can be conned to the relation of total cost and
total revenue. The implications for marginal productivities can
then be indicated.
It is clear that the rm does not of its own volition act so as to
equate average gross revenue to average expenditure, although it
may in the long-run prevent average revenue from being less than
average cost by going out of business.
It is only through the competition of new rms that the demand
curve of the rm may so shift downward as to make the position of
maximum prot one at which total gross revenue equals total
expenditure.
Reserving for later investigation the conditions under which
the demand curve will so shift, let us investigate the implications
of the assumption of zero net revenue. Given
333. = 51, (111)

x x
and
it followsthat the demand curve must be tangential to the total

unit cost curve. . .
In the case of pure competition the demand curve 15a horl
zontal line, and under the conventional assumptions as to the shape
of the cost curve, the tangency will be at a unique pomt, that of
It follows from Euler's generalized theorem on homogeneous functions that the
Hessian of a homogeneous function of the rst order is identically zero.- ObV:::g'
Stable eClllilibriumfor a rm under pure competition is impossxblem these CerllmS a .
minimum average cost. This follows from the fact that average
cost must be equal to marginal cost, and that the latter is rising.
6R 6C C . .
5; = p = ?; = ; = minimumaveragecost." (113)
0f course,
px = ? ww. by assumption, (114)

and
__ _
W: ax so. twa (115)
Hence.
x = ? (pv. (116)
This looks supercially like Euler's theorem, but it is not. For

Eulers theorem is an identity and should be written
x M., (117)
wM:
whereas this is merely a condition of equilibrium holding at a

single output.
Moreover, in the case that the demand curve may be negatively
inclined, we get the more general formulation
R
3: =

OJ
p,, (118)
? "N
This formulation differs from that of Walras, Wicksell, and

Hicks in that the condition of minimum unit cost is derived as a
theorem from the condition that total revenue equal total expendi
" At a minimum of average cost
,
a
_.
C)
__
xC C : ......
!
&: xi ax x :
ie. a10
= and >0
ture. It is this last condition and the forces which lead to it that
are of importance, and not the question of homogeneity at all.
It is quite clear that in the real world net revenue is not zero
for all rms, nor is it tending towards zero. This is true under
pure competition as well as impure competition. It is clear that
this residuum must be due" to something, and it may be labeled
by any name we please (rent to institutional advantage, etc.).
The existence of this residuum does not imply any indeter
minacy whatsoever. Optimal output, revenue, expenditure, and
the difference between these two terms are all fully determined.
Of course, under ideal conditions this residuum will be capitalized
by going concern valuations. Economists, remembering the class
ical solution of the problem of distribution in which the shares of
two or more factors were simultaneously residually determined,
have swung too far in the opposite direction. The attempt to
explain" all residua in terms of marginal productivity analysis
applied to a wider production function can always be done by
convention, but is devoid of empirical content.
It is convenient to have an analytic denition for the case in
which competitive conditions between rms are such that the
demand curve of any rm will always shift until net revenue is
equal to zero. The term free entry" may be dened as the con
dition under which this holds. Of course, this classication cuts
across that of pure or impure competition. Thus dened, free
entry is a condition to be looked for empirically, rather than one
imposed upon the data a priori.
I suspect that part of the economists intuitive desire to dene
a category of prots as distinct from rent to institutional ad
vantage stems from a subconscious remembrance of the old
fashioned distinction between natural" and contrived scarcity.
Perhaps too little is heard today of this distinction, which has
important connotations for social policy and welfare economics.
SUMMARY
In conclusion, a summary formulation of the analysis is pre

sented to replace the famous three Theorems of VValras. These,
aside from being redundant and ambiguous, are not each of the
same order of meaning. The formulation here is for the continuous
case and in terms of marginal equivalences, but the more general
88 FOUNDATIONS OF LCONOMIC ANALYSIS
formulanon terms of marginal mequalxtles readnly suggests Itself

'lhroughout a rm wnth gwen productxon functlon, factor puces
and demand conditions ls pomtcd The general case of pure or
lmpure compctmon ISconsxdered
I The rst fundamental assumption IS that the rm tnes to
maxnmize xts prots and from thus the followmg Internal eondtttons
of equrhbrlum can be deduced
A Any output whneh ISproduced must be produced Withfactor
combinations such that total cost 15a mmlmum As a result of
thts we have two corollanes
1 The margmal product1v1ty of the last dollar must be equal
m every use
2 The puce of each factor of productron must be proportzonal
to marginal phys1cal productlvnty, the factor of proportnonahty
being marginal cost
B That output wxll be selected whxch maxrmlzes net revenue
total cost being opttmally determined by the prevxouscondmons
This unphes
1 The equality of marginal cost and margxnal revenue the
slope of the latter bemg the smaller
2 In combmatnon With the prevnous condltnons under A .
also have the marginal value productwtty of each factor equal to
Its pnce the rst term bemg dened as margmal revenue times
marginal physmal productrvxty
3 Total cost must not exceed total revenue, smce othermsc
the rm would go out of busmess
Il If we Impose by arbltrary assumptlon or hypothesxs the
external conditlons that entry be free 1e , that total revenue be
equal to total cost, then
A Product W111
be exhausted by demtron
B The demand curve must be tangent to the umt cost curve
In the case of pure competltron thlS Implles mlmmum average cost
" If the enterpnse under consrdemtnon owns producuve resources WhIChhave a sale
value on the market It IS necessary that net revenue be at least as large as the sale
(hquxdauon) value of these resources As an Internal eondmon of equzhbnum 1r sale
value of owned resources Alternative uses elsewhere Introduce opportunity costs
Aside from the above general conditions of equilibrium, it has
been shown how the denition of an extremum position may be
utilized (a) to evaluate the direction of change of variables with
respect to parameters (prices) taken as data, regardless of condi
tions of continuity, and (b) to develop reciprocal relations imposed
upon demand derivatives, where these exist.
CHAPTER V
THE PURE THEORY OF CONSUMERS BEHAVIOR
IF ONE were lookIng for a smgle crIterIon by whlch to dIStInguIsh

modern economrc theory from Its classrcal precursors he would
probably dec1de that thIs usto be found In the Introductlon of the
so called subjectlve theory of value Into economIc theory Thls
revolution In thought broke out almost Stmultaneouslyalong three
fronts and wrth It we assocrate the names of Jevons Menger
and \Valras
Moreover, It Is thls part of economIcdoctrme thch has proved
to be the center of so much controversy Indeed many crItIcsof
the orthodox tradItIon have Identied the whole body of economIc
theory wnth the belIef In that abstractIon homo economzcus In
fact, many economIsts well WIthIn the academic fold would sepa
rate economlcs from soc1010gyupon the basrs of ratlonal or ma
tIonal behaVIor where these terms are dened In the penumbra of
utIIIty theory It would seem extremely Important therefore to
know clearly what 15contained In the com entlonal utIIIty analysrs
If only to understand the consequences of denyIng Its valIdIty
EVOLUTION OF UTILITY CONCEPT
The concept of utIhty may be and to have been undergomg

throughout Its entlre hIstory a purgmg out of objectIonable and
sometImes unnecessary, connotations The result has been a much
less objectIonable doctnne but also a less Interestmg one WIth
out domg JustIce to the subject these developments may be sum
marIzed In a brIef way It must be clearly understood however
that these are the movements of the pIoneers of thought TheIr
work appears chIey In academic Journals and has lIttle Inuenced
the general class of economIsts
(a) One clearly delineated dnft In the hterature has been a
steady tendency towards the rejectron of utIlItarnn ethIcal and
welfare connotatIons of the Bentham SIdiIck Edgeworth va
nety These matters stxll recere conSIderatIon In questIons of
90
THEORY OF CONSUMERS BEHA VIOR 91
normative policy, but they are clearly separated from the problem
of consumer's behavior. Although especially marked in regard to
inter-individual welfare comparisons, there is the same tendency
in connection with the analysis of the behavior of a single indi
vidual. Only as obiter dicta do we nd in the modern literature
discussions of particular pleasures as being pure or impure, etc.1
(b) Concomitantly, there has been a shift in emphasis away
from the physiological and psychological hedonistic, introspective
aspects of utility. Originally great importance was attached to
the ability of goods to ll basic biological needs; but in almost
every case this view has undergone extreme modication. At the
same time, there has been a similar movement away from the con
cept of utility as a sensation, as an introspective magnitude. It is
not merely that the modern economist replaces experienced sensa
tion or satisfaction with anticipated sensation, desire, according to
the now familiar distinction between ex post and ex ante analysis.
But much more than this, many writers have ceased to believe in
the existence of any introspective magnitude or quantity of a
cardinal, numerical kind. With this skepticism has come the recog
nition that a cardinal measure of utility is in any case unnecessary;
that only an ordinal preference, involving more or less but not
how much, is required for the analysis of consumer's behavior.
Indeed, so far has the reaction gone that it is the belief of many
that nothing remains but an empty convention. Others, who do
not admit the hollowness of utility, have in some cases embraced
a formulation of the analysis which is meaningless in any opera
tional, empirical sense.2 The result is a curious jargon of dogmatic
precepts.
Thus, the consumers market behavior is explained in terms of
preferences, which are in turn dened only by behavior. The re
sult can very easily be circular, and in many formulations un
doubtedly is. Often nothing more is stated than the conclusion
that people behave as they behave, a theorem which has no em
1The Cambridge tradition is perhaps an exception in this respect, although even
here the change in emphasis is notable.
2Cf. Alan R. Sweezy, The Interpretation of Subjective Value Theory in the
Writings of the Austrian Economists," Reviewof Economic Studies, vol. I, no. 3 (1934),
pp. 176185.
pirical implications, since it contains no hypothesis and is consistent
with all conceivable behavior, while refutable by none.a
Nevertheless, as we shall see, the modern utility theory with all
its qualications is not in a technical sense meaningless It 15a
hypothesis which places denite restrictions upon demand func
tions and price-quantity data, these could be refuted or veried
under ideal observational conditions One should have thought
that these empirical implications would have been the sole end of
the theorists who have concerned themselves with these matters.
Strangely enough, means and ends have been so confused that only
a small fraction of the literature has been concerned with this
problem even indirectly, moreover, in this there are scarcely half
a dozen papers in which valid demand restrictions have been
developed.
I do not propose to defend the fruitfulness of these empirical
restrictions The extent to which they satisfy and unify the fac
tual behavior of consumers cannot be settled by argumentation
However, for better or worse the theory of utility has occupiedan
important position in economic thought for the last half century
This alone makes it desirable that its meaning be clearly understood
PROGRESSION IN MATHEMATICALTnouonr
From the beginning mathematical methods have gured promi
nently in the analysis of utility. Despite the unfavorable reaction
produced among some writers who felt that a spurious precision
was implied by the use of these supposedly exact tools, it is
demonstrable from the literature that symbolic methods have been
an aid to clear thinking and the advancement of the analysis
For those who used this abstract language were forced to formulate
their concepts unambiguously, and so the way was opened for
madiratirmmd.qain im..
It is interesting, therefore, to review very briey the history of
some mathematical aspects of the theory to bring out clearly the
progression in thought through time.
' Still another meaningless ' theory ls held by those writers who speak of behavxor
in terms of the ecanamc prmaple, regardless of whether any empm'l behawor related
to lt ensts . h w,
Thus Edge orth's Mathematical Psychm o'erspenetrating 1n3lghtInto t e we
commonly held m his day
IHEUKY OF CONSUIER'S BEIIA VIOR 93
As early as 1854, Gossen is credited with presenting what is
essentially marginal utility. He assumed this to be a decreasing
linear function of the quantity of any particular good. The utility
function would therefore be as follows:
U = K +(01x1 ' b13512)+ (02x2 _ [723522)

+ ' ' - (1)
Jevons, writing fteen years later, proposed that the utility

function be written as the sum of utilities pertaining to each good
separately.
U = V1(x1) + V2(x2) + + Vn(xn)v (2)
where the functions V.- obey the law of diminishing marginal
utility.5 Specically,
V's/(xi) > O (3)
V;"(x.-) < O.
In his Mathematical Psychics (1881), Edgcworth, going further

than _evons,suggested that the requirement that utility be a sum
of functions pertaining to each good was an unnecessary and indeed
unjustiable assumption: He proposed, therefore, that the utility
function be written in the form
U = 4.9(x1, - - -, x), (4)
where gois any joint function of the quantities of all goods, and
where
.. _ _ai. z () (,,;75 - '
?!) _ axiaxj Z J) (D)
By the end of the nineteenth century many writers, notably

Pareto, had come to the realization that it was an unnecessary and
unwarranted assumption that there even exist utility as a cardinal
magnitude. Since only more or less comparisons are needed for
consumer's behavior and not comparisons of how much more or
less, it is only necessary that there exist an ordinal preference eld.
Walras and Marshall also made the assumption that utility may be written as
in (2). In the case of Marshall, as will be discussed later, it is not clr whether he
rally intended to be taken literally when making the assumption that utilities are inde
pendent, or whether he regarded this as an approximation for small movements under:
certain conditions. "
sProfessor Irving Fisher is also credited with the independent discovery r
possibility at a later date.
For any two combinations of goods respectnvely (x,, ,x,0) and

(x11, . x..), or for brevnty, (X ) and (X'), 1t :s only necessary
that the consumer be able to place them m one of the lollowmg
mutually exclusxvecategories
a (X ) preferred to (X|)
b (X1) preferred to (X ) (6)
c (X ) and (X ) equally preferred or 1nd1'erent
For convenience, we may attach a number to each combmatxon
thrs ls assumed to be a continuous dlfferenttable function Thls
functlon (or rule of numbermg) may be wnttcn
0 = @(X) = 90(1'1. ,x) (7)
It 15so constructed that the followmg three condmons correspond
to the above three respectwely
11 @(X) < AX)
b' 90(X) < (X) (8)
C (X) = 20
a may be demgnated as a utxhty Index The one parameter
famxly of loc1 dened by
50(xl , xi!) : C'
where C IS regarded as a parameter, are desxgnated as mdterence
loc1
It ISclear that any functlon
U = PGP). F'(sa) > 0 (9)
dened by any monotomc transformation of go,xs also a utlhty
Index For
:(X) % an(X) Implles U(X 1) % U(X ) respectively
The converse also holds Thus from any one utnty mac: a1'1
others can be derived by a suitable functional transformation
To summarize, our ordmal preference eld may be wntten
U = FEgacx, , x.):l F'(p) > 0 (10)
where @ISany one cardmal mdec of utrhty
It 15 clear that the choxce of any one numbermg system 01'
utlhty mdex 15arbitral-y The mdr'erence lool are left unchanged
by any alteration of the tags attached to each, provided ordinal
relationships are maintained. In order, therefore, to avoid the
asymmetry of employing any one favored utility index, many
writers (Pareto, W. E. Johnson, Hicks and Allen, et al.) have sug
gested that a notation be employed which is dependent only upon
the invariant elements of the ordinal preference eld, namely, the
indifference loci.
The direction cosines of the tangent plane to an indifference
locus at any point must be in determined ratios. Given any one
utility index, we have
1(ax2 v-0. 6x3 Ugo-I. 6x,,)U_c (11)

as
F'gale'rpz: - - - :F'an.
We may take as given the invariant slope functions
6x . .
()UC : 1R,(xlr ' ' ) x). (1' = 2! ' ' '! n) (12)
These are invariant under any change of utility index, for
lRi _ Ui(xly"')xn) __F,Pi(xly "'!xn)
_ Ul(x17 ' ' ') x11) _ FI01(x11 ' ' : xn)
_Pi(x11"'1xn) -= ...
_ ___SDICCI,
. __, x,.) . (1. 2, , n) (13)
However, if we consider more than two commodities, the func
tions 112cannot be all arbitrarily chosen. In order that there
exist an ordinal preference eld of the type described above they
must, as Professor Fisher has pointed out, satisfy the following
integrability conditions:
IRJJ' _ IR].lRl E IR: _ lRt' IRI (14)
so that the following so-called Pfafan
dxl + 1R2dx2 + - - - + 1.R"d::r:,. (15)
admits of an integrating factor 7(x1, - - -, x), and may be con
verted into the exact differential
dqa = 1!de + (71R2)dx2 + - - - + (71R)dx,.
1dx1 + 2dx2 + + sandxm (16)
96 FOUNDATIONS OI ECOAOHHC ANAL}SIS
where
$91 = 79
and
.2.= (vR) (17)
From this stage it was but one small step to the rejection of
the integrability conditions Thus, Pareto, chLs and Allen and
others h'iie been content to start out With the assumption of a
planar element embodying Indifference directions at each pomt
The latter tuo wntcrs call these the respective marginal rates of
substitution bet een the 1th and rst goods These are written
as in (12), but the functions are not required to satisfy the partnl
differential equations of integrability presented in (14)
1R = .R(x1 , x..) (z = 2 n) (18)
THE DEMANDFU\CTIO\S as A Gou.

\Ve haxe seen an account of the transformations which the
preference eld has undergone through time HO\\CVCf,nothing
has been said as yet as to the use to which these concepts are put
in the explanation of consumer's behaVior This e must nowdo
in order to 1mestigatc the meaningin the technical operational
senseof the various hypotheses
Followmg traditional assumptions of the pure theory of con
sumers behaVior, we conSider a Single idealized consumer buying
goods and serVicesper unit time in a market ii.hose prices he cannot
apprecmbly aect The selling of personal goods and sen icesmay
be at times regarded as negative purchases For present purposes
each good and serVice is taken as clearly dened, homogeneous
diViSible,etc Let us deSignate all goods and semces (ti . x,.)
With respectiie given prices (1); p,.) Total expenditure or
income is dened as
I = up: + rip: + + x..p..= }; W (19)

As of any g en total expenditure and a given set of prices. it is
assumed that our idealized individual \\ ill select some deterinined
amounts of each and exery good (Of course. the amount of some
goods may be zero ) That 15to say, the quantity of each good 15
7Of course Puma. is also an integrating factor
a function of all prices and income.
xl : hl(p11 ' ', pm I)
x2 = thly ' : pm I)
(20)
xn : hn(ph ' ' 'n ? I)

These are the general demand functions. Their derivation is the
whole end and purpose of our analysis of consumers behavior. As
has been reiterated again and again, the utility analysis is meaning
ful only to the extent that it places hypothetical restrictions upon
these demand functions. This is the point of Viewfrom which we
shall proceed.
The Marshallian partial equilibrium demand functions for the
rst good would be, of course,
x1=hl(p11132r ' HIPMi) = D1(pl)1 (21)
where all other prices and income are held constant by ceteris
paribus assumptions. A meaningful restriction upon our price
quantity data would be the hypothesis that an increase in one
goods price will, ceteris paribns, result in a decrease in its quan
tity; i.e.,
6x,
= [Iii < 0. (22)
ap.
Is this derivable from the utility analysis? Can anything be said

about axi/apj, the change in the quantity of one good when some
other price varies? What about xi/I , the rate of change in the
quantity of the ith good with respect to a change in income?
These are the questions whose answers we must seek.
CONDITIONS OF EQUILIBRIUM
It is not necessary that the demand curves be derived for each
of the preference elds dened by (1), (2), (4), and (10), respec
tively. Fortunately, the last includes all previous formulations
as special cases. I start out from the general case of an ordinal
preference eld, later considering the meanings of the special cases.
The utility analysis rests on the fundamental assumption that the
individual confronted with given prices and conned to a given total
expendzture select: that combmahon of goods winch :: hzgheston la:

preference scale This does not reqmre (a) that the mdmdual
behave rationally in any other sense, (b) that he be dehberateand
self-conscxous hxs purchasmg, (c) that there exxst any mtenswe
magmtude which he feels or consults
Our problem. therefore, ls the comparatlvely snrnpleone of
nding a maxxmum for
U = Fcpcl. .x..)] (10)
subject to
>":tmc. = I. (19)
l
where (pl. . pI) are each preassxgned parameters

This IS a constramed max1mum problem, smce equatlon (19),
famlharly termed the budget equation. must be satised Thxs
restrlcts the choxceof quantities Without such a restnctlon the
mdlvndual could presumably purchase an unllmxted amount of
goods up to the pomt of satlanon But m pomt of fact, goodsare
not all free, Wlth a xed Income the more of one good which IS
bought, the less must be consumed of another
In the Appendlx 1t 15shown that we must have as a necessary
condttton for such a oonstramed relatwe maxtmum
U. + AP; = 0, (: == 1, , n) (23)
where X 15a so called Lagrangean undetermmed multnpher Thus
may also be rewritten m enther of the two followmg equzvalent
forms
E: & = 24
UI Px' (2 2v v) ( )
Ol' U
Barge- ___:= x 25)
P1 P2 _ p:- (
Th1s means that m cqunhbnum the ratlo of the marginal utnhttes
of two goods IS equal to the ratxo of thenr pnces, 1e. margmal
utlhtles are proportxonal to prices
It 15clear from the formulation of (24) that lt does not matter
whtch utxhty mdex we use, for
U: = FRO! (! =1v ' ; n) (26)
THEOR Y OF CONSUJlIER'S BEHA VIOR 99
Therefore,
Our conditions of equilibrium yield the same solution, therefore, re

gardless of our choiceof a particular utility index. It is as meaning
less to argue that one particular utility index is really the true
measure of utility as it is to argue that the earth really revolves
about the sun and not vice versa. Only in terms of observations
other than those envisaged in our market place can a cardinal
utility magnitude be dened.
The formulation of (25) gives rise to the familiar interpretation
that in a maximum position the marginal utilities of the last dollars
spent for each and all commodities must be identical. This mag
nitude (= A) has been termed the marginal utility of money,
or better still, the marginal utility of income. It willbe noted that
it is not invariant under a change of the utility index, and so for
an ordinal eld no signicance attaches to its magnitude, nor to
the rates of changes of its magnitude with respect to any variables.
Employing the notation of indifference loci, the same conditions
may be derived. For from (13)
_ Ui(xlv ' ' '! xn)
(i __,2, .. ., n)
IR; _ U1(x1. xn)'
Therefore, the conditions of equilibrium of (24) may be written
IRi(x1,---,xn):=O. (i=2,--,n) (28)

This is the familiar tangency of the budget plane with the indiffer
ence locus passing through the point of equilibrium. Figuratively.
the consumer moves along the budget plane until he attains the
position lying highest on his preference scale, WhiChmust in the
continuous case be a position of tangency; for if the budget plane
crossed the indifference locus, he could advance to a Still higher
position.
We have stated our equations of equilibrium in several different.
but mathematically equivalent ways. The formulation of (23) 15
one in which the symmetry of all variables is maintained, and 50
for deniteness we may concentrate upon it. Our budget
of (19) must
librium can bealso be Satised. and so our full conditions of equi
written
Ui(xi. . x) + A1),= o_ (, = 1, .. .?!)

Pixi + Pax: + - + p,.xn _ I = 0, (29)
These conditions of equilibrium correspond to the set (1) of
the second chapter. We wish to derive from them our demand
functions,
6: = (Pu - .p... I). (30)
which correspond to the equations (2) of the second chapter
Prices and income are regarded as data for this analysis, and we
should like to know how our equilibrium quantities vary iiith
changes in these parameters
Our conditions oi equilibrium are (ii + 1) in number, and in
tolve 2(n + l) unknowns, namely (
A,xl, . tp., - o, pI)
Avoiding now all problems of multiplicities of solution, we may
assume that (n + 1) of our variables may be solved for in terms
Of the remaining (n + 1). In particular (- l\, x., . ,x)may be
each solved for in terms of (p., ,p, I). Hence, we get the
following functions
x. = Mei. .::., I), (a = i, ,n) (30)
and
(_ X) = f(pl! vpI) (31)
Thus, our demand functions can be derived frein our conditions
of eQuilibrium. We have also introduced a new variable (A).
the marginal utility of income, which could, of course, have been
eliminated, but only with a loss of symmetry. Had we employed
one of the other equivalent sets, such as (28), we should have had
n equations between (21: + 1) variables, and so our n quantities
could have been expressed as before in terms of the (n + 1) prices
and income parameters
DISPLACEMENT or EQUILIBRIUM
\Ve have counted our equations and unknowns, and found them
to be equal in number. Subject to certain restrictions, this assures
us that all our equilibrium variables are determined. There is a
temptation to stop at this point and rest content With these
achievements.
THEORY OF CONSUMER'S BEHA VIOR IOI
In view of all that has been said in earlier chapters, it requires

no further argument to show that our task has hardly begun.
There remains the sizable problem of deducing the qualitative
properties of our demand functions from our knowledge of the
properties of our equilibrium maximizing equations.
To do so, we employ the same methods outlined in the second
and third chapters. Let us write the total differential of the equi
librium equations (29)
Udxl + Uiadx2 + ' ' + Uiudxn + id)
= (X)dn (i=1,---,n)
191de + Padx + ' ' + Pndxn
= dl (xidpr + x2dP2 + ' ' + xndPn).
Ol'
U.,dx,-
+ pm = ( mph (2'= 1, n)
1 (32)
2:Padxi: dl " i xde/;.

1 1
These are (22+1) linear equations in the (n +1) unknowns

[d961,---,dx,.,d( M]. Their solution may be indicated as
follows:
: (" UDI'JdPilwI " n:xhdpk)Dn+l.j

dxj = l
n D n (33)
' EZ (" >\)Di.n+ldii|'(dl "' Z dePk)Dn+1,n+1]
d(- A) ___ D 1 ,
where
Un Un Um P1
U21 U22 U2n P:
D _ _U_l
Pi 0!).-| , I. . . . (34)
Unl Un'z ' ' Una Pn
P1 p.) ... p 0
and Da-indicates the cofactor of the element of the z'th row and the
jth column.
The formulae in (33) give the changes in our unknowns {orany

changes in the parameters. prices, and income. As special cases,
the followingpartial derivatives may be evaluated:
% = (_ X)!) "' xtDn4-1}
ai): D (i'j=1! ...,)
gg, = D (35
61 D '
where, of course,
% _ , 1":_ ,
p h. , I _ h; .
Also,
( ) : - U" Dem-1 stn+Im+lj
6?. D '
(36)
( k) = ' Dru-Lu.
61 D
It is convenient to consider &compound term introduced rst
by Slutsky, dened as follows:
,. ax x . .
: 517: xt'i- ("J = 1r "'In) (37)
From (35) by substitution
K,.=(~x)%2. (z',j=l,---,n) (38)

Alsodene
*l: af): +x. ___(7*)]_l_.

_ ?(M a! ("), (:_-1!
_ v) (39)
Ol'
,.=__ (;=1_..._,,) 1w,

The expressions x,/p., Bac,/BI,K,. are all properties of the
demand functions and are empirically determinable under ideal
conditions. We seek restrictions on them. .
Inspection of the determinant D reveals that it is symmetrical
with respect to %and j, since
Us) = U
I'HEUKY UF CONSUMER'S BEHAVIOR 103
Hence,
("' MD? (_ Dia
Kji = D , = D = Kiii (41)
ie,
+;
ax. ax- ax; 6x,-
(. .
(42>
What is the economic interpretation of the compound term
Kit = 3%+ xi % ?
It has been called by Slutsky the residual variability of thejth good
for a compensated change in the ith price.B
This can be made more clear by the following considerations.
Thus far, we have imagined the individual to be maximizing his
utility as of given prices and total expenditure. A little thought
will reveal that utility will be maximized as of a given expenditure
only if the level of utility which is being realized is being achieved
in the cheapest possible way; i.e., expenditure must be minimized
as of any level of utility. If this were not so, the same level could
be achieved with some money left over; this remnant could be
spent to buy more goods, and hence a still higher level of utility
could be attained.
Along any indifference _locusthere exists for any set of prices
an optimum set of purchases which minimizes total expenditure.
That is,
x: = W'Dbi.up. F(so)]. (J' = 1, "un) (43)
For
U = F(o) = constant,
we are conned to the same level of utility. It could be easily
shown that
Ich-=(91)
P. U=constant sw: (i.j=1.---.n) (44)
In words, K5.-is equal to the change in the quantity of the jth
good with respect to the z'th price, where the 1nd1v1dualmoves
' E. Slutsky, Sulla teoria. del bilancio del consumatore," Giamale degli economisli,
LI (1915), 1923.
See chap. iv, pp. 62-65.
104 FOUNDATIONS 0F ECONOMIC ANALYSIS
along the same indifference locus and keeps lus expenditure down
to a mmxmum before and after the change m pnce "'
From the formulatzon of our cqwhbnum equations un (29) we
have seen that the properties of the demand functzons are nal
accted by our choxcc of utlllty Index Thus may be shown et
phcutly from the Identttxes
U=H @@
U: : F'. (46)
Un = [3500 + Fn'P191 (47)
( A) = F'( A'), (48)
where (--A') xs the marginal utlhty of Income for the unhty
Index a Let
DI s: lLli
Pi 0pl, (49)
From (47)
D={1UJ
P: {={Ml
0 m (Al:Peal
0 p; P.!
0
s (Pr'D'
Snmxlarly the following relatnonshnp holds for all cofactors
Do = (JD.; (1 J = 1. . ") (51)
Hence
(- un ___(mon, (52)
D D
so that K,. 13 an luxuriant under any transfer-manon ol utxllty
Index Lemony nothmg :: :mphed jar cmpmca! pme behamr 5?
the chanceof any partmdar uhhty index
Inspection of our ethbrmm condmons m the form of (24)
and (19) reveals that they are unaffected by a proportional change
In all puces and mcomc our equlhbnum values remam Intact for
such '1 change le .
x.=h(p. p.. I)=h'(mp. mp. ml) (:=l, .it) (53)
here m as any posxtwc number Mathematically our demand
For another lntcrprcmt on see H Schultz The The"? ! anrcmml of
Demand (Ch (ago Univers ty of Chicago Press 1933) PP 4345
THEOR I' OF CONSUIER'S BEHA VIOR 105
functions must be homogeneous of order zero. Employing Euler's
theorem for homogeneous functions, we have "
6x,- _x_.- x
a__. ax _
:>-+--+
6911"+p-;b a+
I= O. (z=1,---,n) (S4)
Dividing through by x.-yields the following relationships in terms
of elasticity coefcients:
17i1+m2+''+77in+77u=0. (i=1,---,n) (55)
where
_ % Pi
Tm _ 6171'x.
is the elasticity of the 13thgood with respect to the price of the

jth good, and
_au
77 _ BI x,
is the income elasticity of demand of the 13thgood. Intuitively,

we should expect this equality, since an upward movement in all
prices is equivalent to a decrease in money income.
Thus far our analysis has not been completely devoid of mean
ing. The following empirical restrictions have been found to hold
for the demand functions.
I. They are homogeneous of order zero; i.e., a simultaneous
doubling of all prices and income leaves all quantities demanded
invariant. This implies as we have seen
ax.- c'l__x.- _ E . ___ . __ "6
PIiP1+P2P2+ __l'an?
pn- I I) ('L 1: In) (D)
or in elasticity terms
71i1+17i2+"'+77in=77ili (i=1,--,n) (57)
i.e., the sum of the elasticities of a good with respect to each and
every price is equal in absolute value, but oppOSItem Sign, to the
This may also be proved by substitution from (35)
, DIVii:
EP I=Z(;+%:)*=("M;D 0
by a well known theorem on determinants that the expansion Of the elements r
column with respect to the cofactors of a different column must vanish.
mcome clastxmty of demand for that good These are n restnc

tnons whxch are not compatible With any and all pnce quantlty
behavzor, and so are meaningful "
Because of thus homogenerty condttlon xt :s not necessaryto
take as mdependent vanables the 1: puces and mcome These
(72+ 1) variables may be reduced to n vanables by con51denngthe
ratios between any n and the remmmng varxable
Thus, e may dwlde through by any puce, say the puce o!
the rst good. to get
.]
x-=h(pz. ,p.I)=h*(1,), %],)
.. . ?! 3 & __
8 (131.91. 91.?!) ("'1, ,) (58)
Thxs IS equivalent to settmg the pnce of the rst good equal to
umty, and usmg xt as our numerotre
However, a more symmetncal measure suggests ntself Dmd
mg through by I, we get
xi=h'(pll npmn=h'(%:%| '.pII)
H(I'l
_ 3 &: I&) (1= 1, ,n)()59
From our def nmon of meome or total expenditure as
I - ZM
:
we have the {allowtng(n + l) restrictions on ehsucmes of demand
Z I! : = 1
l
and
.
: &u = - h
I
where
x
7
l
13the proportion of mcome spent On the :th good However these are not mailing;
restnctwns state they are consequences of our denmon At best they C
reveal that we have not applied our dened operations With numenall accuracy
Let
a,
u
NI?
Q n r" : V
Hence,
xi = Hlah ', a"). (60)
The as here are very natural units to employ, since they involve
only the dimensions of the respective quantities. In words, a,-may
be dened as the proportion of total income required to purchase
a single unit of the jth good.
II. We have also the following reciprocal integrability con
ditions.
x- x- ax.- ax.- . .
Kj'=P+xii=;+xi=Ki (11.7=11"'1n) (61)
i.e., the residual variability of the jth good for a compensated
change in the ith price is identically equal to the corresponding
term for the ith good with respect to the jth price. These are
n(n 1)/2 independent meaningful conditions.13
MEANINGFUL THEOREMS
Thus far almost nothing has been said about the directionsof
change in our equilibrium quantities of goods demanded with re
spect to changes in prices and income. Does the utility anaIySis
have nothing to say upon this question? The answer can be sought
along lines indicated in previous chapters.
Before clouding the air with determinants, let us make a
common sense appraisal of the situation to see whether we cannot
suggest a simple answer.
First, suppose the individual to be constrained to move along
the same indifference locus. Let him be confronted With a set of
prices, and attempt to attain this level of utility" in the cheapest
They hold, of course, only for the individual demand functions. Moreover, they
reect differential properties of our demand functions which are .hard to Visualizeand
hard to refute. For our empirical data consists of isolated pomts. _The must be
smoothed in some sense before our relations an be tested; the smoothing. even by the
best known statistical methods, is to a degree arbitrary, and so refutation and verication
are difficult. _ _ . ..
I have tried, but thus far with no success, to deduce implications of our mtegmblllt
Conditions which can be expressed in nite form, i.e., be conceivably refutable merely
by a nite number of point observations.
possrble way Consrder the set of puces (91, , p..) There
wrll correspond to this an optimal set of quantltles (n, ,xg)
such that total expendrture rs as low as possrble : e ,
? P.x. ;? P.xu (62)

where (X) 15any other pomt on the locus
F[(X)] = F[(X)] (63)
Consrder now a second set of puces, (Pr. , Pn). and the
correspondmg optimal set of goods, (x11, xn), lying on the
same mdrfference locus as the rst Then
was. s pers (64)

l I
In equatlons (62) and (64) any values of x (along the same locus)
may be Inserted the respective left-hand srdes In particular,
we may write them respectwely
m.! >:pm. (65)
I 1
and
m. 2":px. (66)
I. l
This means that the optimal set of goods for each respectzveset of
pnces cannot cost more than the other set of goods (optimal fora
dlerent set of pnces)
Rewrrtmg the equations, we get
p.(x. x.!) s 0, (67)
1
Mac.1 x.) : 0 (68)
Add these two equatlons to get

i (pr1
_.. P\)(le _ xaa) g 0 (69)
I
If the two equxhbnum pomts are assumed to be always drstmcc.

and If a proper absolute mmrmum xsassumed to be always realize ,
THEORY OF CONSUMERS BEIIA VIOR 109
then the equality sign may be dropped, and this may be rewritten
21:(P3 "' p,)(x _ xi") < 0, (70)

This may also be written
21 ApiAx; < 0, not all Ap; = O. (71)
Suppose we allow but one price to change, say the kth; then
all but one term of (71) vanishes and we have
AdQ-Apt. < 0; (k = 1, ' , 71) (72)
i.e., as the kth price increases, all other prices being held constant, less
will be bought of the kth good. It must be emphasized that this
holds only for a movement along the same indifference locus, i.e.,
for a compensated change in price, and does not mean that with a
givenmoney income a change in one price will necessarily result in a
decreased amount taken of the corresponding commodity. It will
be noted that the above proof does not involve the calculus at all;
usingonly the operations of addition and subtraction, the denition
of a maximum position may be utilized to derive meaningful nite
demand restrictions.
Employing only the most elementary logical and arithmetical
operations, we can advance matters still further. Consider any
initial set of prices and income (1210, ', Pn, I)- Corresponding
to this there will exist one or more optimal sets of goods. Select
one of these and designate it by (x1, ', xn0)' Consider now a
second set of prices and income (1211,- ' ', Pnl, Il), and a correspond
ing Optimal set of goods (x11, - - ,x,.l).
Let us consider what would have been the cost of the second
batch of goods at the prices of the rst. This will be
P103611+ P2x21 + . . . + pnoxnl = ? pix.-1. (73)
If this cost is equal to or less than the amount of money that the
rst batch actually cost, we have conclusive evidence that the
second batch is not higher on the individuals preference scale than
the rst batch; for if it were, the individual could not have been
in equilibrium in the rst place, since he would not be mlnimizmg
total expenditure for the attained level of satisfaction. In other
IIO FOUNDATIONS OF ECONOMIC ANALYSIS
words, ii he could have bought the second batch, and he bought

the rst. we rule out the possibility that he prefers the secondto
the rst.
Our theorem is
lpxJ wxf implicsFMXUJ FMX?) (74)

More Specically,
$1512fo< )::pximplicsmam < Foam]. (75)

Similarly,
>::p.'x.< me wouldimplymom < more]. (76)

It is obvious that
i px.< p.x.. (m
and I l
% p.x. < 5:51;st (78)

cannot both hold simultaneously. for this would imply
: mam] < mum]. (79)

mumu > mom. (80)
which is a contradiction.
This gives us a condition which holds for any movements. not
merely for compensated ones.
Equations (77) and (78) may be written
"2 p.(x.' x.) < 0 implies f; paw x,") < o, (si)
1 1
br
f: 9.41:.< 0 impliesif (p. + %. < 0. (82)
! 1
If we assume that our demand functions are single-vained.and

: to consider only distinct points, this may be broadened to
.LJJ..|.;V4\Jur uvrvouman o DLH}! VJUK III
the following form:
):. PiAxi % 0 implies 1 (p,- + Aps)Ax.-< 0_ (83)
The importance of this result can hardly be overemphasized.

In this simple formula are contained almost all the meaningful
empirical implications of the whole pure theory of consumer's
choice. Moreover, these are expressed in the form which is most
suitable for empirical verication. So fundamental is this condi
tion that (as I have shown elsewhere) it provides a foundation for
the theory of economic index numbers and the utility analysis, and
affords the most convenient path for the derivation of all known
restrictions upon the individual and general demand functions.
From this condition alone can be derived the followingrestric
tions upon the demand functions:
(a) They must be single-valued; i.e., to each set of prices and
income there corresponds a unique set of goods.
(b) They must be homogeneous of order zero; i.e., a change in
all prices and income in the same proportion must leave all quan
tities unchanged. All the properties of condition I of the previous
section must therefore hold.
(c) All known valid qualitative restrictions upon the slopes of
the demand functions. These will be indicated below.
Elsewhere " I have suggested as new foundations for the pure
theory of consumers behavior the conditions (a) and (b) and
equations (83). At that time I did not realize that (a) and (b)
were redundant in the sense that they themselves could be deduced
as theorems from the assumption of (83) alone. In other words,
this single condition provides us with complete foundations for the
theory (with the reservation concerning integrability). .
The proof of (a) and (b) as theorems can be indicated Simul
taneously. Consider an initial price and income situation (pf, - - -,
71
1). Corresponding there is a set of goods (x1, ' ' '. x110)
Suppose now all prices and income to be multiplied by the same
The only point upon which this formulation does not throw light is that of in
tegl'ability. Even here, a proof may still be forthcoming by Whichthis condition may
be slightly generalized to include the question of integrability.
A Note on the Pure Theory of Consumer's BehaViOI'."Economica, February,
1938. pp. 61-71.
112 FOUNDATION!) OF ECONOMIC ANALlSIb
posmve quantnty. m. (mp1. , mpu", ml) Correspondmg there

Is a second set of quantmes (xxl, , xn) We wushto prme that
the second batch of goods xs Identlcal, commodxty for commodlty
With the rst
By hypotheSIS,
I = mI (3.1)
Hence,
;??st : m px (85)
Also,
N = "2.0. (2 = 1. .n) (86)
Therefore,
onxtl : Pools" (87)

: 1
But also
$: Nx. = 5i pm (88)
1 1
But this IS&contradiction, for condltxon (83) says that
Zn:pJxJ = X":p,x\ Implles 22:p.x. > px. (89)

I 1 I I
Therefore, these cannot be two dzstmct pomts Hence.

xl, = 37.0 (7 = 1: In) (90)
For m = 1, condition (a) ts deduced as a specml case of (b)
Thus far, we have made no use of the calculus We may now
proceed to denve condltlons upon the vanous partial denvatwcs
of our demand functrons
Let us go through a lxmutmg process and wnte (83) 111the
followmg differential form
igdpx. < 0, (91)

for
)::pdx. ==0,
lot all do.. or dp, ==0
TIIEOR Y 0F CONSUMER'S BEIIA VIOR I I3
In this expression the dx's and the dps are differentials, not
innitesimal increments.
Regarding prices and income as independent variables, from
our demand functions of (20) we have
dx.- = >":hjdp;
1
+ ;;de (i = 1, n) (92)
But for p,:ixj = O,

1
dl : 2i:[dei + Z;depJ : ? depJ (93)

Therefore,
dx. = 01; + x,h,')dp, = K.,dp,. (94)

1 !
Equation (91) becomes
2": (h,-"+ x,h,)dp.dp, ; o, (95)

! rM:
OI

i z (mama.- g 0. (96)
1 1
This is a negative semi-denite form; semi-denite because for all

prices changing in the same proportion, it vanishes due to the
homogeneity condition.
This is also derivable in at least two other ways. From
equation (71)
i AP.Ax.- 0 along an indi'erence locus (97)

1
This may be written
{:! ddx. < 0 along an indierence locus.

(not all differentials vanishing). (98)
II4 FOUNDATIONS OF ECONOMIC ANALYSIS
But along an ind'erence locus, from (43)
xt : lbEplv' ' 'r pm F()Jr (i = 11 ' ' " n) (99)

where W is homogeneous of order zero in the 9's. Also
dx. = zjgfdp, ( = 1, ,n) (100)

Therefore,
::: >:Wap, % o. (101)
But, of course, for a movement along an indifference locus. ie..

for a compensated price change,
x,
ap.5 $}. E Kip (102)
Therefore, (101) may be written
)5 K.,dp.dp, g 0. (103)
l !
Finally, in the Mathematical Appendix the followingalgebraic

theorem is stated: Let
[A"] = [%] : [AN]! (in].: 1. "" 71) (104)

be the rst 71by 7:matrix of the inverse matrix of [D] Then
>":2":A.,th, ; o (105)
1 I
because [D] is the matrix of a negative denite form under C"

straint. Now from (38)
K,, = (_ 70A. (105)
Since (-A) > 0. our theorem again follows. This is a direct
algebraic proof of our theorem. .
The meaning of the requirement that the form m (96) be nega
Jauv1m1 \IJ VUIVUUJVIDIXODLH/l VIUI IIS
tive semi-denite may be briey indicated.16 Let
Ku K12 K
K21 Ku K2,,
K = [Kn-I = _ _ _ - (107)
Knl Kn2 ''' Kan

Then because of the semi-deniteness
|K| _=_o and i Kim-p, = 0. (108)

1 ]
However, the principal minors beginning with the rst alternate in

sign from negative to positive, i.e.,
Ku K12 K11 K12 K13

IK11l< 0; K K > 0; Km K22 K23 < 0, etc. (109)
21 22 K31 K32 K33
Thus, the following demand restrictions are implied.

6x.- 6x,- ._
'Pl-xiaI<O, ('1) in) (110)
(6p.+xal)(ap,-+xiaI) apj+xlal >0

(1:11.: 11 ' '! n)) (1:J)r etc' (111)
Condition (110) was rst developed by W. E. Johnson and

Eugen Slutsky, presumably independently. We see that it is not
possible to deduce that
% 112
ap.- < 0, ( >
the ordinary expression for the law of demand. For if axe/6.1

be sufciently negative, axi/ap. can be algebraically positive. Th1:
is the phenomenon alluded to in the well-known Giffen s Paradox.
" If integrability is not assumed and Kg,-;6 Kia. the exposition may be easly
modied by the substitution throughout of the term (K.-,-+ Kid/2- _
" It is only by making additional, and demonstrably arbitrary, .assumptions that
various writers have been able to derive the so-callcd law of diminishing demand.
116 FOUNDATIONS OI" ECONOMIC ANALYSIS
The assumption that the form in (96) be symmetrical and nega

tive semi-denite completely exhausts the empirical implications
of the utility analysis All other demand restrictions can be de
rived as theorems from this single assumption These are bold
statements, but they are substantiated by the fact that it is passable
to work backwards from the assumption of (96) to an integrable
preference eld displaying the properties necessary fora maximum"
CONCLUSION
We have come a long way in this chapter. Despite its lofty

beginnings, the pure theory of consumer's behavror, u.hen its em
pirical meaning is nally distilled from it. turns out to be one Simple
hypothesis on price and quantity behavior. This may be written
imp. + Manx. < 0.

for
E {hdt; 0, not all Ax; = 0,

1
subject to the qualications indicated above. Alternatitcly, in

may u rite this as
I l an 0x;
set-w
-
) P mg 0,
3. ~ d .d
" Only a sketch of the prooi ol this statement need be gwen Wme
:. - !!!: val)! ( . le Un)
_ an Ox; ax. ' 95:
A"-3-p-,+357-1(Ba, a:-'n).
Dene a new set of variables
o " f(a|| ! au). or d'. - Filu c1)
such that
" _ ]IEpth ,n). . Pal. IS.)] : Gwh op.)!
and a;
0.1" , _ = _:
65;5 G @ l,! = A! . C! a
Then there exists a inaction
P - (Bl I n)
or
9 " @(xh 01.0)
u hich satises the properties of our preference eld
THEOR Y OF CONSUMER'S BEHA VIOR I I7
where the equality sign holds only for all prices changing in the
same proportion.
Many writers have held the utility analysis to be an integral
and important part of economic theory. Some have even sought
to employ its applicability as a test criterion by which economics
might be separated from the other social sciences. Nevertheless,
I wonder how much economic theory would be changed if either
of the two conditions above were found to be empirically untrue.
I suspect, very little.
A NOTE ON THE DEMAND FOR MONEY
One special problem in the theory of value has been touched

upon at various places in the previous discussions, namely, the
value of money itself. Probably more has been written upon this
subject than upon any other in economics, and most of the issues
raised are not germane to the present investigation. However, it
is a fair question as to the relationship between the demand for
money and the ordinal preference elds met in utility theory. In
this connection, I have reference to none of the tenuous concepts
of money, as a numeraire commodity, or as a composite commodity,
but to money proper, the distinguishing features of which are its
indirect usefulness, not for its own sake but for what it can buy,
its conventional acceptability, its not being used up" by use,
etc., etc.
The most interesting problems arising in connection with money
are linked up with the fact of uncertainty in the most general
sense, leading into liquidity considerations which cannot be dis
cussed here. However, it should be possible to indicate in a few
pages the way out of certain false dilemmas connected with the
demand for money for so-called transactions purposes.19
On the one hand, there are writers like Mises who would explain
the value of money in marginal utility terms such as might. be
applicable to any commodity; on the other hand, to economists
like Schumpeter the peculiarity of money lies in its essential lack of
direct usefulness and in the fact that its value is not capable of
explanation in the usual utility terms. On the whole the latter
"' For a summary of the extensive Continental speculations on these matters sale
Howard S. Ellis, German Monetary Theory, 19051933(Cambndge, Mass., 1934), Part
vxew IS the least rmsleadmg, but, as Walras showed many years

ago. 1t 15p0551b1eto modrfy utrltty analysrs so as to take account
of the pecuhar properties of money The latter. who above all
others developed the notlon of general eqmltbrmm m Wthh all
magmtudes are Simultaneously determmcd by efcacrous mter
dependent relatlons was able to remam undlsturbed by the fears
of literary wnters that there was something vmously mrcularm
assummg the exrstence of prices and of a value for money the
midst of the process by Wthh that value was to be determined
Today after the recent contributions of Keynes 1t ISpartlcularly
reuardmg to go back to retarmne the elaborate dlscusswnof
hquldlty preference, encazsse deszree. etc , m Walras The latter
was so SOphtsttcated as to have outgrown the quantity equation m
later edltrons of hlS work, although he contmued to beheve 1n\\hat
lStoday called the "quantnty theory" , very nghtly, ln my opmlon
he 1neffect reversed the commonly met dictum that the quanttty
theory should be scrapped but the quantrty equation ISuselul
Here I shall only consnder the demand for the holdmg of money
by the consumer As before, ordmal utrhty or preferencedepends
upon all commodmes but the (n + 1)th good, M . wxllbe taken to
be money, Wthh erIds benet only m Its ultlmately being gnen
up Possesswn of an average amount of 1t y1elds convenience m
permlttmg the consumer to take advantage of offers of sale, 1n
{aerhtatmg exchanges, m brxdgmg the gap between receipt of m
come and expenditure, etc The average balance 1sboth used and
at the same tune not used, xt revolves but rs not depleted. Its just
bemg there to meet contmgencxes lS valuable even 1f the con
tmgencres do not materlahze, ex post Possessnonof this balance
then yields a real servzce, whxch can be compared wnth the duect
utlhttes from the consumptzon of sugar, tobacco, etc , m the sense
that there 15 some margm at whxch the 1nd1v1dualwould be 1n
different between havmg more tobacco and "lesso't a cash 'Da'lan.
wrth all of the mconvemence which the latter condxtxon1mplles
But there IS thls dlerence Gwen physxcal amounts of to
bacco, food, ballet, etc have sngnzcance 111terms of the ant
pattern of the consumer, but It 15not possnble to attach Slmar
s1gmcance to a gwen number of physmal umts of money. say to
a number of ounces of gold It would be otherwxse 1nthe case of
woldwhich was to be used to ll teeth. but such uses of money 1n
THEOR Y OF CONSUJIIER'S BEHA VIOR I I9
the industrial arts we purposely neglect. The amount of money
which is needed depends upon the work that is to be done, which
in turn depends upon the prices of all goods in terms of gold.
The above remarks are by now so well known as to appear trite
and trivial. But let us translate them into mathematics. Our
ordinal utility is now a function, not alone of the physical quan
tities of goods, but it has within it prices as well. This is a serious,
signicant alteration, for, as we shall see, the empirical properties
of the demand functions are changed by this innovation. This is
not the only case in which economists have found it necessary to
introduce prices into the indifference loci; there is also the example
of goods which have snob appeal, or scarcity appeal, which are
valued for their exclusiveness so that preference for them is altered
by changes in their relative prices. This Veblenesque effect need
not detain us here.
Our utility function will be of the form
U(x11 ' '! xny lm, pl: ' ', Pn)
E U(xlv "'! x117MpmyXP], "',n)
F(xl,...xn,.l.,...,E
III M) (113)
where the function is homogeneous of order zero in all of the prices,

so that a doubling of all prices (including the price of gold) at the
same time that quantities are unchanged will leave ordinal utility
unchanged. This stems from the hypothesis, which need not be
true in the short run or under particular expectations, that money
is evaluated only in terms of the work which it has to do. It will
be noted that I have not set the price of money equal to one.
Actually, we shall avoid confusion in thought if we refrain from
doing so. Any other commodity may be used as numeraire, or we
may express prices in any units we like. Of course, it will still be
possible to speak of the gold price of things, and after we have
learned to do without the use of the money commodity gold as
numeraire, we can then fall back upon the simple convention of
expressing prices in terms of it. Of course no reader will think
that I attach any particular importance to goldor any other metal;
any conventional unit which serves as money will do.
Walras was careful to pomt out another Important dustmcnon
consumptxon of goods 13a ow per umt tune, so much tobacco per
year etc , but the gold balance 15an Inventory or stock We may
speak of 1ts prxce two senses, the prnce of gold compared to the
prices of other thmgs as one ounce of gold IS worth tuo beaxcr
skms whxle a pound of tobacco nsworth three beaver skms, or an
ounce of gold rs worth two anythmgs while a pound of tobacco
ls worth three anythmgs In our notatnon this ns? But no
may also speak of the puce of the use of gold per umt txme In a
capxtal market where peOplecan both borrow and lend at a gnen
rate of mterest, thls pnce 15necessanly related to the rate ol
terest Thts IS true even if the mdmdual 1n questron does not
hax e to stay m debt for the amount of hrs Cashbalance,1n any case
there ts the opportumty cost of holding money m the senseof the
Interest he mnght have earned by lending this sum
If we w1sh to exclude dynamlcal consrderatnons from our (llS
cussnon as much as posszble, the Simplest assumptron would seem
to be that the 1nd1v1dualmaxrmrzes the above expression sub)ect
to the followmg budget eQUatlon
Ax, + rpmM = I (114)

I
where pnces, mcome, and Interest 7' are given to the mdmdual
Thus, the amounts pend out (or foregone) m every perlod for the
use of money are treated as subtractions from Incomeavailable for
expendxture upon consumers goods The condrtlons of equl
lrbrlum are exactly as In chapter v equation (29), except that the
margmal utdmes of the goods are affected by the level of pnccs
dlrectly, and we now have an addrtronal unknown, M, to be de
termmed But we also have an add1t1onalequatton
E _ 115)
6M + xpr 0 (
Thus, the use of money 1sbought up to the pomt where 1tsmarginal

utility (convenience) 1s proportxonal to ltS cost, or the mtlcrest
charges Wthh must be pard fonts use The secondary mequa IthS
are exactly as m the usual case 1e , the prrnc1pal rumors of t e
THEORY OF CONSUMER'S BEHAVIOR 121
bordered Hessian must oscillate in sign, with money treated as an
(n + 1)th good.
From the conditions of equilibrium our demand curves are as
follows,
xi=hi(P1,",Pn,PmyL7)y (1:=1,",7Z)
M = MUM, ...,pm p,I, r). (116)
These equations are homogeneous of order zero in all prices and

income just as in the usual case of demand. With the interest
rate unchanged, doubling all prices (including the price of gold)
and income will leave all quantities unchanged. However, if we
double all prices and income in terms of the price of gold, then
with r unchanged, the demand for physical quantities of money
will be exactly doubled. Mathematically .M is a homogeneous
function of order one in terms of all other prices and income. This
is the peculiar feature of money. Another way of saying the same
thing is the assertion that the demand for money in terms of its
own price, all other prices held constant, is necessarily of unitary
elasticity. This is a familiar proposition in the history of the
quantity theory. The reader will note that this is a meaningful,
refutable hypothesis which is capable of being tested under ideal
observational conditions.20
The further qualitative properties of the demand schedules can
be determined by differentiation of our equilibrium equations.
However, they will not be so simple as those in the usual demand
case, because in varying any one price, we shift every equilibrium
equation through the direct inuence of each price on marginal
utilities. Still, income elasticities behave more or less the same as
before; also the demand for money in terms of the interest rate
behaves just like the demand for any good in terms of its own price
under the usual utility assumptions. Thus, if interest is raised,
and at the same time income is increased so as to leave the indi
vidual equally well off, then the amount of money demanded must
fall. If money is not an inferior good, and we should expect it not
to be, then an increase in interest will lower the amount of money
demanded even for an uncompensated change.
Jacob Marschak, Money Illusion and Demand Analysis," Reviewof Economic
Statistics, February, 1943, pp. 4048.
The complete results of such dIfferentIatIons are gwen below

m partItIoned matrix form "
h": hp; hu hr;
AI; III ill; M,

J
" '1 l ' a
LU aU. w.
Unit 611! Pu Edi-A6 "gbm 0 0
= _ 1% 62H azU (117)

6!!! aM= 9" aMap, aMapmH " "
Pk Pm 0 _; x, r! -1 p,..M
We may leave as an exercnseto the Interested reader to xenly

thxs and to worl. out the Interestmg and specnal case where the
ordmal preference eld takes the Simple form U(x1 x.. {AM/D
QUALIFICATIONS INTRODUCED BY UNCERTAINTY

The above analysrs Of the demand for money connes Itself to
the so called transactlon aspects The really Interestmg problems
anse when uncertamty as well as transaction friction Is admitted
) the PICture Some of the best lmes ever wntten m the con
nectIon are those to be found In Keyness General Theary and m
chkss Value and Capzlal Space can be devoted to only one com
ment here In the paSt the structure of Interest rates and assets
has been SUJCCtto only perfunctory notlce so that the recent
analySIs In terms of quuIdIty preference 18 of the greatest rm
portance But It would be a mlstake whrch would be as damaging
to further analysrs of quu1d1typreference as It would be to classent
doctnnes, If It were thought that uncertainty and liqmdlty drffer
enttals are the sme qua non for the CXISteCB
of a rate of Interest
" Reference may be made here to the valuable contnbuuon of C E V Leser T139
Jmers Demand for Money Econometrxa vol XI no 2 (April 1943) PP 125-
111LUK1 ur bUNbUMEK'S BEHAVIOR 123
Such a view can be compared with a theory of land rent based upon
differences in the quality of different kinds of land. I believe that
the analogy is not a supercial one.
Specically I think it would be misleading to say that interest
is simply the price of money; it would be quite possible to have an
economy in which money did not exist, and in which there still was
a substantial rate of interest.22 Nor can I agree with all of the
formulations of Hickss brilliant chapter xiiion Interest and Money,
such as . . . it must be the trouble of making transactions which
explains the short rate of interest (p. 165);" . . . if bills are
are perfectly safe, [there would be] no reason for the bills to fall to a
discount (p. 165); Securities which are not generally acceptable
in payment of debts bear some interest because they are imper
fectly money (p. 167).
It is true that in a world involving no transaction friction and
no uncertainty, there would be no reason for a spread between the
yield on any two assets, and hence there would be no difference in
the yield on money and on securities. Hicks concludes, therefore,
that securities will not bear interest but will accommodate them
selves to the yield on money. It is equally possible and more
illuminating to suppose that under these conditions money adjusts
itself to the yield of securities. In fact, in such a world securities
themselves would circulate as money and be acceptable in trans
actions; demand bank deposits would bear interest, just as they
often did in this country in the period of the twenties. And if
money could not make the adjustment, as in the case of metal
counters which Aristotle tells us are barren, it would pass out of
use, wither away and die, become a free good.
In the preface to the reissue of Risk, Uncertainty and Prot 2
Frank Knight makes the penetrating observation that under the
conditions envisaged above the velocity of circulation would be
come innite and so would the price level. This is perhaps an
Elsewhere I have developed this notion briey under the title The Rate of
Interest under Ideal Conditions," Quarterly Journal of Economics, LlII (February,
1939), 286297.
In a world without uncertainty where money has the same yield as other assets,
its velocity becomes indeterminate. Note that this is the case when the interest rate
is zero.
Risk, Uncertainty and Prot (London School of Economics and Political Science:
Series of Reprints of Srce Tracts, no. 16, 1933), p. xxii.
I24 FOUNDATIONS OF ECONOMIC ANALYSIS
over dram'mc way of saymg that nobody would hold money and
1t would become a free good to go mto the category of shellsand
other things which once served as money We should expect too
that rt would not only pass out of crrculatlon but rt wouldceaseto
be used as a conventional numeralre ln terms of Wthll pnces are
expressed Interest heal-mg money would emerge
Of course the above does not happen m real lnfe precrselybe
cause uncertainty contmgency needs non synchromzatnon of
revenues and outlay transactlon fnctrons etc etc all are W1thus
But the abstract Specralcase analyzed above should warn us agamst
the facrle assumptron that the average levels of the structure of
Interest rates are determined solely or primarily by these dlffer
entra] factors At tlmes they are primary and at other times such
as the twentles thls country they may not be As a generalxza
tron I should hazard the hypothesrs that they are hkely to be of
great xmportance m an economy m whtch there rs a quasr zero
rate of Interest I thmk by this hypothems one can explam many
of the anomahes of the Umted States money market m the thntles
A further defect of the reasoning that I have been crrtrcrzmgles
In the tendency to regard as universal the hypothe51sthat Interest
yreld 15mversely correlated wrth an asset s nearness to money so
that normally long rates are above short rates This 15Of m
'uccord thh much of economic history for reasons whrch I do not
think are difcult to elucndate 5
" See D Durand Bas c Y elds of Corporate BOnds 1900-1942 Techn l Paper 3
016: York Nat onal Bureau of Eoonomc Research 1942) Also F A Lutz The
Structure of Interest Rates Quarterly Journal of Econonuc: LV (1940) 36-63
CHAPTER VI
TRANSFORMATIONS, COMPOSITE COMMODITIES,

AND RATIONING
LOGARITHMIC TRANSFORMATIONS AND ELASTICITIES
THROUGHTHE INFLUENCEof Alfred Marshall economists have de

veloped a fondness for certain dimensionless expressions called
elasticity coefcients. On the whole, it appears that their impor
tance is not very great except possibly as mental exercises for be
ginning students.1 As we have seen, most of the laws of eco
nomics are qualitative and ordinal rather than quantitative, and
where such are concerned the problem of dimensions is of no
consequence.
Furthermore, while elasticity expressions are invariant under
changes of scale, they are not invariant under changes of origin.
Since there are no natural zeros from which we measure economic
magnitudes, the elasticity expressions can be seen to be essentially
arbitrary. Thus we meet in economic analysis such concepts as
exports, net purchases, amount of factors supplied, etc., all of
which are differences measured from arbitrary bases.'-
Mathematically an elasticity expression between two magni
tudes, such as price and quantity, consists simply of the logarithm
of one of these quantities differentiated with respect to the log
arithm of the other. Thus,
112,,= (dx/dp)(p/x) = (1log x/d logp ==Bac/Ep."
1There is perhaps some usefulness of the concept of elasticity of demand as giving
an indication of the qualitative behavior of total revenue, but even this is only the
consequence of neglecting to dl with total revenue directly.
2This was realized, somewhat inadequately, by Wicksteed who denied the validity
of the concept of supply, preferring to use the notion of reserve demand"; only he
went to the opposite extreme in attaching sacredness to his reformulation.
A calculus of operations with elasticity terms can be devised on the basis of the
properties of logarithmic .derivatives. See R. G. D. Allen, Mathematical Analysis for
Economists (London, 1937), chap. x, particularly the reference there to the work of
D. G. Champernowne.
125
Not only are elasticity expressions more or less useless,butin
more complicated systems they become an actual nuisance,con
verting symmetrical expressions into asymmetrical ones,and hiding
the deniteness of quadratic forms. This could be illustratedby
utility analysis, but can be amply demonstrated by the slightly
simpler case of prot maximization, where constraints do not enter
the picture. Let us consider a rm buying inputs (,.. ,v,)
perfectly competitive markets at given prices (m,, ~ -, u,.) With
the demand curve for its product and its production functionbeing
known, total revenue becomes a determinate function of the inputs
bought. Total expenditure being dened as the sum of the
amounts paid to all the factors of production, it is clear that the
lirm's profit can be written in the form
:r = Roll, : Un) "' 20,0). (1)

1
For prots to be at a regular maximum, we must have

121(in i vn) _ w; = 0! (3 = 1' I n)
[R.,] negative denite. (2)
Suppose \ie are interested in the inverse demand functions
giving the amount of each factor of production which willbe bought
at each set of prices. By explicit differentiation of our equilibrium
conditions above, we find
[it]
aw, = [R.,]I. <3)
Since the inverse of a symmetrical negative definite matrix is itself
Actually it is a little misleading to say that an elasticity expression is necessarily
without dimenSion " For take any absolute derivative,such as dr/dp, which is cer
tainly not dimensionless, involvmg as it does the dimensions [output times output dwidcd
by value] Even though it has dimenSions. it is still the elastiCity of same expression
Thus if
= = KP).
and
y = 5(9).
here
y = "i q =":
then
E_y dr.
Eq'd
TRANSFUKMA 'I'IONS, COMMODITIES, RATIONING I27
symmetrical and negative denite, the complete conditions on the
demand functions are summarized in the above expression.
If one were interested in the corresponding elasticity coefcients
[Evg/ij], one could as an afterthought introducesuitable factors
into the matrix of equation (3). This usual procedure has been
termed by Professor Lange the indirect method.5 He suggests
an alternative direct" method. I should like to point out a third
method, which carries out the Lange procedure to its logical con
clusion. Before doing so, however, I should like to express the
opinion that in this matter indirectness is a virtue rather than
a drawback.
It will simplify the discussion if we adopt a mnemonic notation
whereby the Jacobian of one set of variables with respect to another
is written in a form reminiscent of that for a single derivative; i.e.,
. _ % _ d_3'
J(y11'.')y!xh"'1xn)-[axj:|dx' (4)
The reader may then verify the identities
1342 _ l
dydx _ dx
5
d_x__
dy avr
_ dx ( )
which parallel exactly those for single derivatives.
In terms of this notation the Jacobians of the transformations
Vi = 10 711', 2),- = el'i _
W.- = 10:20.. w.- = e"'-' (* = 1, n) (6)
can be written as
@. 'U
_[2] = [Wk]
dv _ Ui ,
W 5," w
FL%l
% = [217]- aw = Wi
where S,-,-
is the Kronecker delta, equal to one for like subscripts
and vanishing for all others.
Osr Lange, Theoretical Derivation of Elasticities of Demand and Supply: The
Direct Method." Ecanometrica, X (1942), 193214.
Then It follows that
dv dVdDdu 5: _
m = W= [Elm] 'WJ 8)
Tlns Is the so called 1nd1rcctmethod The nge direct method
mvolvcs computmg the same matrn by the followmgcquahty
dV __ dW 1 v,R., 1
an? [W] = [r] (9)
There seems to be no specxal advantage m thmsprocedure on
the other hand there rs the great d151dvantage of loss of symmetry
prior to mvcrszon and a cloakmg of the demteness properttes of
R and Its Inverse Of course. what IS coxercd up can later be
unscrambled. but a good deal of waste motton would seem to be
Involved
However If e wxsh to carry the direct method to Its nal con
clusmn 1t would seem logtcal to replace the v's and 205 m the
ongmal prot crpressnon by theu equwalences m terms of the
variables V and W \thh the latter being gwen, the rm would
vary the Vs so as to mawmlze prots The condxtxonsof equr
hbrmm would be
S.(V1, V,.) eweI ==0, (z = l, , n) (10)
and
[SU _ v.w..,]: H
negative demte In the above expressxons xt xsto be understood
that
S(V1. . V..) = R(e"'. . e). (11)
and that subscnpts stand as usual for dl'erentxatron
If we dterenttate our equlhbrrum condmons we easdy derive
the 1dent1ty dV
.E'JV
=-..H'[v,w,,,] (12)
where Hl xssymmetncal and negatlve demte Thus except for

column factors the elastlcnty matnx ISsymmetrrcal By du 1d1ng
through wnth these factors and demng K . ==Mo./): vim...e
easnly arrive at the famzhar xdentlty
Kirin : Km,. (13)
' H Schultz Theory and Measurement of Demand (Change 1938) chap mx
1 [(111voru1<MA 1 1UIVJ, COMMODITIES, RA TIONING 129
Also we may verify the relations
"h'i "7t'1'
"if < Or
vii 7m l > 0, etc. (14)
Relations (13) and (14) are also immediately derivable from the
indirect method, but not easily from the Lange method.
If one must use dimensionless" expressions, it would seem de
sirable to replace the usual elasticity expression by the more sym
metrical coefcient
62).- \l'wvj
Mg,
: aw,- w.- (15)
The reader can easily verify that symmetry and deniteness are
preserved in all matrices, that these coefcients are pure numbers,
which in the case of diagonal elements reduce to the old denition
of elasticity.
I should like to conclude this discussion of elasticity expressions
with a warning that the transformations break down when any of
the variables are zero or change sign. In earlier sections it has
been shown that productive services or inputs can be regarded
simply as negative outputs. These can be converted into log
arithms only after a reversal of sign. This would offer no problems
were it not for the fact that many of the magnitudes are neither
always positive nor always negative. As an illustration of the
difculties which may arise, consider the classical case where the
total of all expenditure is taken as zero. Here the partial deriva
tives of the various goods with respect to hypothetical income
changes have a denite meaning, whereas their elasticities cannot
be dened.
GENERAL TRANSFORMATION OF INDEPENDENT VARIABLES
The logarithmic transformation discussed in the previous sec

tion is but one special case of the transformation of our inde
pendent variables. It is desirable to examine the behavior of our
equilibrium conditions under general transformation of our inde
pendent variables because in economics as in other disciplines we
wish to free ourselves from any particular scheme of reference, .or
coordinate space, in favor of generalized cobrdz'nates. This desire
is not only aesthetic but is necessitated by the fact that in real life
I30 FOUNDATIONS OF ECONOMIC ANALISIS
ne do not always meet naturally dened commodntres In busmess
cycle theory we often use the concept of the cost of lmng " taken
to be the pmce of some composxte market basket of commodltzes
But even If we conne our attentlon to what 18ordmarrly calleda
commodxty, such as nheat, we nd ourselves deahng With 1
composrte commodity made up of wmter wheat sprmg wheat of
varymg grades Each of these m turn 1s a composne of hetero
geneous components, and so forth m mmte regressron
There 18a related reason {or analyzmg the effects of a general
transformation Among the many recent advances contamedm
Professor] R chkss Valueand Capztalperhaps the most impor
tant from an analytlcal pomt of v1ew was the enuncnatronof the
prmcxple that a grOUpof commodltres has the property of a smgle
commodlty If their prices all change m the same proportron This
theorem had been partially antxcxpated by Professor Leontref7but
It \\ as Professor Hicks who made xt the cornerstone of his exposx
tron Ex ery mathematical economlst, workmg m many vanables
nds dlfculty m explalmng hrs theones m words or dragrams By
means of the Links Leontref theorem all but one of the vanables
can be lumped together as a smgle \arnable, and a tolerably
Simple hterary explanation can be phrased We shall see how thls
theorem emerges as a specxal case of general theorems concerrung
transformations
The most general condmons of ethbnum are of the type rst
encountered m chapter n equation (1), wh1ch we may rewnte m
matrvr form wrth suppressed subscnpts as
f(x, a) = 0 (16)
If we subject the Independent variables to the non smgular
transformation
x == T(x). (17)
whose Iacobxan
= [T.-]
never vamshes then our equlhbnum condrtrons become
f[T(-'b') 0!] = Kw. a) = 0 (13)
"W Leonuef Composrte Commodlt es and the Problem of Index Numbers
ztnca vol IV (I936)
TRANSFORMATIONS, COMMODITIES, RATIONING I31
If the relationship between the equilibrium values of our un
knowns and the parameters was given by
x = g(a). (19)
the new relationship will be
f = a(). (20)
which will be equivalent to
T (5) = g(a)- (21)
Most importantly, it can be easily veried that
dx dx di:
=' @)
We may turn directly to the general extremum problem where
some quantity is to be at a relative maximum or minimum with
respect to ?: independent variables, which are themselves under
m constraints. In matrix notation we have
z = f(x) (23)
G(x) = 0.
At a regular maximum it is necessary that the matrix
Us. GJ (24)
be of rank m, and that
h'Huh (25)
be negative denite subject to
hG, = 0.
where
H,, = Cf.-...,,
+ {gnaw}. (26)
where the Asare Lagrangean multipliers.
After transformation we have
&, Ja [% '[f,,c;,]. (27)

' See Mathematical Appendix A, Section III.
where prlme on a matrix [a ,] Indicates 1ts transpose [0,1

Smce the rst matnx on the ght hand Slde 15non smgular the
rst order rank condntnonsare mvanant under transformatlon It
15also easy to show that
h r = h'IInh (2%)
where
11G, E h'G, E O
dx
h Ex.] h
so that the second order comhtlons are Imanant under trans
formatxon
Altematxvely tl'us mxght have been proved by showmg the m
vanance of the relevant bordered determmants consequenceof
the relation
1/3; 0, Q: 0 II,, G, d_x 0

dz: dz, (29)
G, 0 0 I G, 0 0 I
It WI" bc noted that the Lagrangean multlphers (A. >)
are Invariant under transformations of Independent xarnbles
There are economic as well as mathematncal explanations of thus
If we Insert parameters (a; an) m our equations above sothat
our constramts become
G"(x) ou; = O (30)
then one can usdy \errfy b) dlrect differentiationthat

dz
_ A (31)
[dat ]o herasco mt [ 3
Earlier we have dem ed specml cases of this m connection mth the
marginal utlhty of Income (chapter v p 100}and m'zrgma'tcost
(chapter n p 65)
The above rehtxonshsp can be (km ed heunsttcally by d1cr
entzatmg II With respect to a; holdmg !" other varnbles constant
As 1th other opentlons 1molvxngLogrmgcan multtphus this Is
m analytic truck whose Justncanon hes :ts equu "dmc: With
expressmns whteh c he rigorously dem ul Ivy more round nhout
methods
Explicit attention is called to the fact that transformation of
the independent variables of an extremum problem subjects the
matrix of the relevant quadratic forms to a cogredient congruent
transformation which leaves the deniteness of the quadratic form
unchanged, and also leaves the signs of principal minors of a given
order intact. This is in consequence of the equivalence of our
equilibrium equations to an extremum condition, and need not be
true for other types of equilibrium.
Thus, when Hicks 9sets down as a condition of perfect stability
under general market exchange the requirement that all the
principal minors of a matrix
- ["]
Pi
be positive, we have as yet no way of knowing whether or not this

is independent of the particular coordinate system chosen. The
same may be said of his concept of imperfect stability." The
reason for the ambiguity lies in the inapplicability of the proofs
given above to asymmetrical matrices.lo
This section has dealt with the invariance of an extremum posi
tion under transformation of independent variables. We turn now
to the problem of transformation of the dependent variable.
TRANSFORMATION OF DEPENDENT VARIABLE
It is by now well known that the equations of demand are

independent of the particular choice of utility index;11 i.e., all of
their properties are invariant under a general transformation
U = FG). (32)
where the latter is a one to one, always increasingfunction. For

SimPlicity we may assume that it is twice differentiable. .
In the most general case where f(x) is to be made a max1mum
subject to one or more constraints on the xs given by
G(x) = o, (33)
' J. R. Hicks. Value and Capital (London, 1939). chap. v and Mathemt AP
pendix, pp. 315-317. .
"This problem is discussed again later in this chapter.
See chap. v, pp. 94, 99, and 104.
the equxhbrmm conditions are completely expressed by the condn

tron that [}}sz be of certain rank, and that the quadratzcform
h'Huh be of certain demteness for values of h satisfying
h'Gz = 0 1 (34)
Now If we subject z and the Gs to the transformatton
F = F(z) and E = E(G), (3a)
and ask that the new varlable F be mammlzed subject to the new
constraints our condltlons Involve the rank of [F:Ez] and the
demteness of h'Ruh, subject to
"'-E: = 0. (36)
when.
R:: : [Fu + 'E mkEUk] (37)

1
By actual dl'ercntlatnon we easxly had the relationships between

the new and old matnces, namely
' 0
[ms] = U,G,][ F((f) 3(0)] (38)
Slnce the last of these matrices 15a dlagonal matnx thCh 15t be
non Singular, the rst order rank properties are clearly ma
under this transformanon, as ISthe actual solution m x use
We may 51m1larlyvenfy by dlfferentlatlon that
MR): = 111177,, + ): mkEg'(G")G.,']h

!
+ h'EF'w, + i m,,Emfafjh (39)

|.
Prom the rst order rank condltlons there must be lmcar de

pendence between the f. and Gz columns of preCISCY the 5 to
make the second part of the right hand slde Ofthe aboxe equation
vamsh for adrmssnble h's From the monotonie property Of the
functlons F and E xt follows that [F'Eg'] are non vamshmg, hence
the transformed quadrant: form retains 1ts demteness
" See Mathematrcal Appendix A Secuon V
TRANSFORM'A TIONS, COM'MODITIES, RATIONING I35
This completes a sketchy proof of the invariance of our equi
librium conditions. This whole problem is of interest primarily in
connection with utility and consumers choice theory, but our
theorem applies just as well to the maximization of prot. If we
maximize the square of prot, or the log of prot, or 90 per cent of
prot, or that monotonic function of prot which the Treasury
tells us we can retain as net income, there will be no change in our
optimum price and output policy. In connection with the produc
tion function the combining of the factors of production so as to
produce a given output most cheaply could be done as well if we
worked with some indicator of output rather than with output
itself; alternatively, we might work with the iso-product contours
alone, taken without regard to their numbering. However, it is
important to realize that the variables, prot and output, are
measurable under ideal conditions; consequently it is articial and
unnecessary to replace them by a transformed variable.
Explicit evaluation of the new Lagrangian multipliers shows
that they have been changed by the transformation as follows:
F'UML
mb= W (40)
The signicance of this for the marginal utility of income appears
in the discussion in chapter vii.
TRANSFORMATION OF PRICES
Thus far I have considered transformations of the unknown
variables, leaving the parameters, a, unchanged. We are able'to
evaluate changes of the new variables with respect to a by equation
Can those who have a fondness for mrdinal utility nd support for their position
by analogy to other disciplines, such as phys? In my opinion, CF?tam physical
concepts, such as temperature, were for a long time treated by the PhYSCIStSfs Prdmal
magnitudes. Their behavior could be described by any one Ofmany di'erent "d'Fators
whose relationship to each other was not linear. Finally it was deemed convenient t
give conventional, privileged position to one such index, namely that based upon the
properties of a perfect gas." Nor is the use of the word force" in PhYSlCSbk5 ? gd
precedent for utility analysis; much of physics can disPcnse With the term, and. any
case the things for which it stands, such as rate of Changeof momentum, or derivative
of a potential function, n be given strict observational values, which are not .mvanant
under monotonic transformation. Perhaps a better precedent for economicsls that of
statistics, where likelihood or its logarithm is used indierentl) depending upon nu
merl convenience, and without metaphysl connotations.
I36 FOUNDATIONS OF ECONOMIC ANALlSIS
(22) However, m the new coordinate system we are no longer

able to state denite theorems regardlng the Signsof these changes
In the old system we could demtely state that
av!
62171
< 0, (4|)
but our new coefcrent (a/aw!) lS a hnear combination of terms

of the form (am/am), the Sign of whnchcan be anythmg
And yet, since there was nothmg pnvrleged about the rst
coordmate system, It 18mamfest that there must be somewherem
the new coordmate system theorems no less denlte than tn the
old As we shall see thts hunch" 15correct There are such
theorems, and they are to be derived by con51dermgthe changes
of our new variables w1th respect to specxed composztechangesm
our parameters That IS at the same mme that we subject our
equilibrium unknowns to a transformation. we subject our param
eters to a transformatton The latter transformatron rs not the
same as the former, but, as wrll be shown presently, the transforma
tron of parameters IS related m a denite manner to the trans
formatzon of quantlttes
For the present purpose, rt IS not necessary to ecammethe
problem ln full generahty We may consxder the Important case
of a rm mammrzmg prot as m the first section of th1s chapter
The case of an extremum under constraint, such as appears m pro
duction and utrllty theory, can be developed m a Straightforward
manner by the reader
Let prots be written m the form
(v.w) = Rel, ,a) 22m, (42)
where the 20'sare parameters taken as gwen by the rm, and the
v's are productlve mputs The vsand w s are Just specral casesof
:: s and 19s,havmg spec1c reference to Inputs
lf now we dene new commodities (vl, , z ), we ought to be
able to nd demte relatrons between thur changes and changes
m the" prlces, Just as we hate been able to do m the old coordinate
system But what are a to mean by the pram: of the new com
modztzes, (ml, , w) Only one answer 15 p0551ble '15 must bc
obvmus to anyone who has worked wrth the price of a market
basket of goods
TRANSFORMATIONS, COMMODITIES, RATIONING 137
The price of the latter is reckoned as a weighted sum of the
individual prices, where the weights arc chosen so that the total
valueof the new commodity, equal to ma, is the same as the value
of the component parts. This suggests a general rule to be fol
lowed. Throughout all transformations, value magnitudes are to
be conserved; i.e.,
z wkv. = >":@ (43)

1 l
regardless of the transformation. For this is a value magnitude

(dollars, pounds, etc.), which is not arbitrary once prices and
quantities in any coordinate system are given. On the other
hand, prices and quantities involve the particular referencesystem
being used, and are thus arbitrary.
It is a familiar mathematical problem to subject two sets of
variables to transformations, now assumed to be linear, so as to
conserve their inner products as in equation (43). Variables
related in this way are called cantragredient variables.15
If we subject the quantities 2)to the non-singular linear trans
formation
1! = c, 27 = 611), (44)
then
wv = 11/55. (45)
If, and only if,
10' = 'n'/rl, or u)= cw, (46)
will (43) be realized. Thus, the related transformations of prices
and quantities are
2) = cz), z;= clv, (47)
w = 61'11'), 1b = C'w,
where the transposition and inversion of the matrices is to be noted,

along with the fact that the relationship between the two trans
formations is reexive.
This problem arises in higher geometry, in many branches of mathematics and
in multiple regression analysis where predicted dePendent variables are to be independent
of transformations of the predicting variables. As the latter .undergo a linear tranhs
formation, the regression coefcients must be adjusted accordingly 50,35 to leave t c
predictions unchanged.
" M. Bocher, Introduction to Higher Algebra, p. 108'
After transformation prots become
new) =- m; w) &:EU) I M,. (48)
In form this is the same as the original; consequently, it follows

from the maximization conditions that everything that couldbe
said about (dv/da) can be said about (d/dan). Thus,
d
= (in!): (49)
is symmetrical and negative denite.
This might also h. ve been deduced from the {act that
d_5_ ddvdw
a. = 532723 (5)
Substituting into this the equations of (47), we nd
d _ dv ,
35 = c 1317:c . (SI)
It is a classical mathematical fact that a matrix's propertiesof

negative deniteness and symmetry are preserved alter a con
gment transformation of the above type.
There are many equilibrium systems encountered in economies
which do not arise from an extremum problem and which cannot
be converted into this form. The various simplied versionsof
the Keynesian system provide but a rst example of what is after
all the general case. A second example is that of the genera}equi
lzbmmz equations of Walras. An intermediate case is a third elf
ample in which we assume consistent preference behavior, inni
tesimal indifference loci at every point subject to certain generalized
convexities, but without the assumption that integrability" con
ditions are satised so that the local" indifferenceplanar elements
can be joined together to form a family of indifference lof!
Thus, our preference eld may be characterized by a dl'erentlal
Mathemanl Appendix A, p 368 The rst proof above an be used to demieI
proof for thus
Inverseof classical
a denite theorem
matnx, Also,denite.
a, is Itself if we set This
c = norelated
[a], It 13easy
to theto"an
show thatP")
t:
function or price potential" discussed m chapter m, p 56 The latter In mm new
related to the solledLegendre transformation See A Wintner, Analytical Founda
hon: of Celestial Mechanics (Oxford 1941). chap I
expression, or Pfaian, for which no integrating factor can be
found. Only in the case of two goods can we always nd an
integrating factor which will lead to an exact differential" which
can be integrated."
The rst of the above three examples is of no interest in this
connection since we cannot pair a set of price" variables as
conjugate" variables to quantities."
Turning to the second example of general equilibrium, I
shouldlike to point out a grave diiculty with the Hicksian concept
of stability of general exchange and production. In later chapters
it is argued that stability is an essentially dynamic concept, and
that the formal statical treatment of Hicks leads to conditions
which are neither necessary nor sufcient from a dynamical point
of view. Here it will be argued that the Hicks stability conditions
are unsatisfactory from a purely statical point of view.
This may be illustrated by the somewhat simplied system of
chapter v of Value and Capital, whose equilibrium is given by a
set of equations of the form
x"(p1, ...,pn) = O. (k = 1, ---.n) (52)
To be perfectly stable the ]acobian, [dx/tip], must have all its
principal minors oscillate in sign. A matrix with this property I
shall call Hicksz'an. To be imperfectly stable, we only impose the
weaker condition that the (n 2) and (n 1) order minors be of
the opposite sign.18
Since no particular denition of commodities or frame of refer
ence is fundamental, if the Hicks conditions are fundamental, they
must be invariant under transformations of price and quantity.
A single example suices to show that a matrix a which 15HICkSlan
becomesunder transformation c[dx/dp]c1, and this need not be
Hicksian, nor even imperfectly stable.
" The problem of integrability goes back to the "classl" discussions'pf Irving;
Fisher and Pareto. In their A Reconsideration of the Theory Of Value, Parts
and II, Economica, XIV (1934), 52-76, 196-219, Hicks and Allen touch upon It. bu:
not in an altogether satisfactory fashion. The most illuminating treatment is that; of
Gcue-Roegen, "The Pure Theory of Consumer's Behavior," Quarterly Jounm o_d
Economics, L (1935-36), 545-593. See also the discussion of chap. Y, P,; 95 above, Env
P' Samuelson, "A Note on the Pure Theory of Consumer's Behawor, Economica.
(1933). 6171,353354,for some unsolved problems in this connection
" Valueand Capital, pp. 67, 315.
I40 FOUADATIONS OF ECONOVIC ANALISIS
Thus for
if _' [
dp l
0 10]
1
_ [ 2l
61_ 21 ] (53)
c get
dx 96 96
?? = o 96] (54)
Much ISneither perfectly stable nor Imperfectly stable
Of course :( [dx/dja] had been symmetncal and demte (as
when Income effects can be neglected) thus could not haxc hap
pened but then our system would have been com ertlble Into a
maximum problem
I have shown else here that convexity non mtcgrablc cases
lmphes that the symmetncal part' " of certain matnces be
denite This follons m consequence of the relation
z": Acc./sp. ; o (55)

l
u.here the p s and the xs may or may not be under constramts

By gomg to the hunt c nd that [dx/dt) + dx/dp'J/Z and
fdp/dx + dp/dx'J/Z must be non posxtwe
It xsnot hard to show that thts property, \\ thll we shall call
quan-dcmteness IS Invariant under redemtlon of commodmes
or under a general non smgular congruent transformation Thus
let [a + a']/2 be negatne denite Then the quadratlc farm
h'ah & h' Bij24.4!h + 0 (56)

usnegatne demte Conversely, nfh'ah nsnegative denite, then
a. IS negatwe qua51 demte
Wlth the and of a non Singular matnc b c subject a to a con
gruent transformatlon so that
a. = MI!) (57)
" Every matnx a can be decomposed mm a symmetnml and an asymmemcal
part Thus
= a + a +a -a
2 2
TRANSFORMATIONS, COJUJUODITIES, RA TIONING I41
Then
Ii = [li'bJaEbli] = h'ah, (58)
and the extreme right-hand member is always negative, as must
be the left-hand side. Thus, [(2+ ']/2 is negative denite, and
is negative quasi-denite.
The above analysis may be of interest even where there is in
tegrability. If income effects cannot be neglected, the Jacobians
of general exchange equilibrium need not be symmetrical. How
ever, if the income eectsare symmetrical and denite, or if their
symmetrical parts are denite, or if their symmetrical parts are not
so lacking in deniteness as to outweigh the deniteness of the
substitution effect, then the Jacobians will be quasi-denite. Not
only is this property preserved under transformation, but it is
possible to establish the Theorem: If a is quasi-denite, it is
necessarily Hicksian. but not conversely.
This important theorem can be proved in a variety of ways.
One is to write each principal minor of a in a Taylors expansion
around [a + a]/2 so that the skew elements (a,-,- a,.)/2
appear
in a power series. Then it can be shown that all of these elements
enter in even powers so that the complete a matrix is more
Hicksian" than its symmetrical part. This conrms my previous
remark that asymmetry, per se, makes for greater rather than less
stability. The real signicance of quasi-deniteness can only be
come apparent from the dynamical discussions of stability in
later chapters.
DEMAND FOR A GROUP OF COMMODITIES
Value and Capital will take its place in history along with the
classic works of Cournot, VValras,Pareto, and Marshall. Like the
latter, Hicks has succeeded in keeping formidable mathematical
analysis below the surface of things and locked up in appendices,
thereby securing for his work a much wider audience than would
otherwise be possible. This tour de force was made pOSSIble1n
considerable degree by the repeated use of the already mentioned
theorem relating to the demand for a group of commodities when
their prices all change in the same proportion. We are now 1n a
position to derive this theorem rigorously in a more general form.
142 I'OUNDATIONS OF ECONOMIC ANALlSIb
When a group of puces say (1)1 1)) all move together
nt ISnatural to dene a new commodity ::; by the relatxon
$1 = 91361+ P2352+ + P x (59)
Strxctly Speaking we are not thereby enabled to replace 1: com

modztlcs by (n -r + 1) commodltles We are merely changmg
our frame of reference and so we must lose no dlmenswns or degrees
of freedom m the process Our complete transformation of quan
tltles can be gn en by
:; __ [0PI I[ Pc pI 0 O] x (60)
Here we haxe Simply dened all commodltnes but one to be nden
ncally the same as before the rst commodnty has been replaced
by the new composxtecommodity The contravanant puce van
able must then satlsfy the equatnon
1 O 0 '
-22
Pn
I p (61)
O'-'
0
so as to leave 2 1n: = Z px Even though (n 1) commodltles
are the same as before their pnces must have changed This
wxll not seem strange 11'It IS remembered that a pnce IS not a
property of a good Itself Independently of the frame of reference
mvolved In fact 1t ls qmte possxblefor e1therpr1ces or quantmes
to Lecome negative although ):, pr wdl conserve 1ts ongmal Sign
Now If we change prices (p, p ) m the same prOportlon ).
holdmg constant (p H p,.) It can be shown by exphcxt dr'er
Th 5 s only one of an nn te number of pass ble tmnsformat ons wh ch w ll serve
present purpose
entiation, d/d), that in the new price-quantity coordinates only
the 131is changed with all other 13sremaining constant. Conse
quently, if this price changes, and if income is changed in such a
way as to keep utility constant, then
:15: _
(TI-97:,
= Ku < 0. (62)
Also
Kquz _ 122> 0, etc., (63)
so that all of the properties of the demand functions given in the
previous chapter are satised.
Thus, the demand or a group whose relative prices are unchanged
satises the same inequalities as in the case of a single good.
This is the fundamental Hicksian theorem.21 It can easily be
broadened so that we may work with s(_<= n) composite goods.
Their substitution terms form a negative denite quadratic form.
This is a very useful theorem. It states that any group of
commodities whose relative prices remain unchanged can be lumped
together into a single commodity, and that indifference curves in
fewer dimensions can be constructed which will have all the usual
properties of indifference curves. Thus, all goods but one can be
lumped into one commodity which Hicks calls money," and then
concave indifference curves can be drawn between the good in
question and money.
It would seem that this involves rather a strained use of the
term money," and one almost certain to lead to confusion.23
Very properly when Hicks later comesto discuss monetary matters,
he rejects this earlier notion.
" Proved in his Mathematical Appendix, pp. 311312,as a consequence of his sixth
rule relating to substitution terms. All six of these rules are contained in the statement
h' [X.-51his non-positive denite quadratic form of rank (n - 1). 'hCh vamshes for
values of h proportional to prices, this being an immediate consequence of the primary
conditions of equilibrium and the secondary conditions guaranteeing an ettremum
under constraint.
Value and Capital, p. 33, passim, diagram on P- 39- 5150see A' G: Hart, Peculi
arities of Indifference Maps Involving Money," Reviewof EconomicStudies, VIII (1941).
126.213258imh
an instance is provided by a writer who interprets Hicks literally and con
{usedly attempts to throw light on the "inflationary gap" by means Ofan indifference
diagram in which money is taken as a mtchall for all but one commodity. M. W. Reder,
Welfare Economics and Rationing." Quarterly Journal of Econonncs, vol. LVII (1942).
144 FOUNDATIONS or ECONOIUCANAL} SIS
Tm: GENERALPnonuzu OF CouposrrE on
AGGREGATECoxmoomas
Any economtc system hen wen ed carefully mll be found to
consrst of an almost uncountably large number of \arnbles It IS
almost a necessrty If ad.ances 1nanalysns are to be made to sumphfy
matters artxctally so as to reduce the number of \arnables whlch
are to be handled T111515done m a number of different mays
Some authors retreat to '1one or mo commodity \\ orld order to
deme more precnseresults the penalty for this hes the difculty
m estabhslung the relation betuecn the sumphed construct and
complex reality Thls ho ever ISat least an honest procedure
Other \xrxters 13hto hate thexr cake and eat It too to work mth
only a few variables and at the same tune retam an "nr of reahsm
and versxmlhtude
Here too a \anet} of artful dodges are open to the mvestngator
(1) He may hold other thmgs equal (2) He may concentrate on a
representatwe rm or family, each of \\ hom IS donng about what
all are domg (3) Or he may \\ ork u 1th certain aggregate, com
posrte magnitudes such as bales of output, "socnall) necessary'
labor age units " cost of hvmg. real natnonal output. etc
lt ns the tlurd deuce whlch 18 of Interest here There :s
nothmg 1ntr1n51callyreprehenszble tn \sorlung \uth such aggregate
concepts On the contrary, abstractron from compleuty xs a
necessary thought process And m any case the most genenl
equlhbnum set up must necessarily stop short of the full total of
all possrble economic \anahles But It 15Important to reahze the
hmxtanons of these aggregates and to analyze the nature of therr
constructlon
Broadly speaking \\ e replace '1number of \anables by a smgle
vanable under two drametrrcally opposuteconditions In the rst,
the variables haxe each the same effect (except 130551nyfor scale
drflerences nhlch dlsappear upon redefimtron) upon all of the eco
nomtc functions under discuss1on In tlns case they can bc
summed and treated as one Thus, we might dene as the same
commodltles all goods Inch h'ue exactly the same Influence on
consumption preferentx elds and production functlons \\ luch are
so to speak 1nmtelysubstxtutable ln practice ngtd adherence to
perfect substxtutabahty zmght lead to the unmanageable result that
; tno things are qwte the same Consequently, goods which do
TRANSFORMA TIONS, COMMODITIES, R '1TIONING I45
not signicantly (for the purpose at hand) differ are treated as
identical.
Actually the above case of substitutability is but one example
of the more general mathematical theorem whereby the identical
vanishing of a Jacobian, or of all of its minors of a given Order,
implies that there exists functional relationships between sets of
the variables, so that many may be dropped out of the picture
completely. It is clear from the implicit function theory that such
simplications are possible only in the case where the original
system was indeterminate. This may or may not be a matter of
concern. What if the variables which are extraneous cannot be
given determinate values by the system of equations which dene
equilibrium? The indeterminate variables may be a matter of
indierence to the economist. Thus, take any determinate system
of economic goods. Let each unit of one type of good be marked
with an invisible serial number, and now let us ask how many odd
numbered units will be bought by a given consumer. Clearly the
answer is indeterminate, but also of no possible interest. If con
sumers did have preferences as between different serial numbers,
then the indifference schedules would be affected by this fact, and
additional equations would be available to determine the nal allo
cation. And notice that even in the case of complete indifference
the nal allocation is not really indeterminate on any particular
occasion. It is just that the determining factors, which are taken
as due to chance by the economist, would be of a different
character.
Unlike the rst condition under which variables are lumped
together because they are innitely substitutable, at the other ex
treme we combine variables which bear an invariant, perfectly
joint" relationship to each other. A notable example is provided
by the classical dose of labor and capital applied to land. Here,
too, the economist customarily relaxes the rigid requirement of
perfect collinearity in favor of an approximate condition. _ThuS,
the most primitive justication for a price level constructols pro
vided by the indisputable fact that prices do generally rlse and
fall in about the same proportions.
" This must be qualied. Our original equilibrium equations may be imbedded
in a still wider set of equations so that the total is determinate, but the original subst
taken by itself is not.
Geometncally thls dn'ers from the rst case m which 1nd1ffer
ence cun es or ISOquants are straight lmes Here the contours are
broken lunesmeetmg at rlght angles m such a way that the goods
or factors wnllby obvxous ch01cebe combmcd together 1n the gwen
proportlons almost regardless of price ratlos Unlike the chksran
bundle of goods the physxcal composmon Is the same regardless of
relatwe puces
Of all composxte magmtudes perhaps the most mterestmg to
the theonst xsthat of an index of the cost of hvmg, or of an index
of consumptnon output Such an mdecISdesxgned to meet certam
specxal requlrements and need not be satisfactory for other pur
poses In partxcular, lt need not represent the deszderatum from
the standpomt of the questions whlch Jevons and other proneers
m the use of Index numbers sought to answer Yet the theory of
these 1nd1cesISof some mterest for 1125
own sake and because ln the
course of mvestngatxon of these Index numbers econormsts mad
vertently stumbled upon ccrtam ordmal relationships whlch are
basm to welfare economics and to consxstent consumers behavmr
THE ECONouIc THEORY or INDEX NUMBERS

Economlsts such as ]evons, Edgeworth Marshall, Allyn Young,
Warren Persons, Irvmg Fisher. Edwm Fnckey, and others have
made contributions to what may be called the statistical theory of
1ndex numbers But what has come to be called the economlc
theory of 11'1denumbers 15concerned wrth qutte dlsttnct matters
To thts theory many economlsts have made contnbutlons A
partial hst would mclude the names of chksell, Konus, Bortkte
wrcz. Bowley, Haberler, ngou, Keynes, Staehle, Leont1ef, Allen,
Lerner, Frlsch. and Wald
Wnth the exceptlon of Leontlel all of these wrlters are unduly
preoccupied With the problem of the price mdex number, mstead of
bemg occupled wnth the more fundamental questlon to whxeh xt xs
only a partxal, and somewhat arb1trary. ans er Thefundamental
problem upon winch all the analyszs rests as that of determmmg merely
from {mat and quantrty data winch of tua summons :5 hzgher up on
Reference may be made to the survey article by R Fnsch Annual Survey Of
General Economic Theory The Problems of Index Numbers Economemca IV (1936).
138also to the amcle by W Leonuef the same Issue The early volumes of the
of EconomzcStudm may be consulted for other dzscussnons
cm individuals preference scale. This problem admits of a partial
answer if certain rigid assumptions are fullled.
It is assumed that there is only one individual, whose tastes do
not change in the period under consideration; or if there are more
than one, that their tastes are identical. It is not necessary to
assume that the regime of commodities is the same in the two
situations provided we follow the convention of setting at innity
the price of any good which is unavailable. At whatever the
prices named, the individual is presumed to buy as many or as
few of the goods as he wishes. This rules out such phenomena as
rationing or monopsony. Ordinarily, a comparison is made be
tween two situations which differ in time, but we might just as well
compare two situations different in any respect, as in the case of
comparisons in the cost of living between two regions.
Our fundamental data consist of prices and commodities, in
cluding productive services which can be treated as negative com
modities, in the two situations, represented respectively by (P, X)
and (Pb, Xb), where these are shorthand notations for the prices
and quantities of n goods. Of course, if we know completely the
individuals whole preference eld, we may simply insert into it
the two amounts of commodities and read off which is the better,
or whether they are indifferent. Since we do not know the eld,
our problem is to go as far as we can on what we do know.
If for the moment we restrict ourselves to the quantity data,
we ordinarily cannot say which of the two situations is the better.
However, in the exceptional circumstance in which one of the
situations contains more of some of the goods than the other, and
not less of any of the other goods, then it is clear which is the better.
Taking situation A as our reference point, and thinking of the
simpler two-dimensional case, we can denitely divide up the whole
of the commodity space into four regions with respect to X by
letting two perpendicular horizontal and vertical lines (not shown
in the gure) intersect at this point. Treating this as the origin,
we can denitely classify all points in the northeast quadrant as
better than X, and all points in the southwest quadrant as de
nitely worse. We are left then with ignorance concerning the in
terior points of the other two quadrants. In these more of one
commodity has been acquired at the expense of less of another, and
until c have further Inform'mon concermng the preference eld
this :s as far as we can go
But lt should be possnblc to utlhzc our price mformatlon If
the 1nd1v1dualISsupposed to be mammzmg lllSordmal preference
eld subject to the gwen budget relatton we may be sure that the
pomt X ISbetter (or not orse) than any pomt between the budget
RECON OF IGNORANCE N ANALYSS
OF NDEX NUMBERS
TY
MMOG
scoono
1|
FMS? COMMIY
FIGURE 1
locus and the axes For all such pomts were available to the m
d1v1dual m the mltlal Sltuatlon and he nevertheless preferred to
select X It follows that for all such pomts
U(X) : U(X) (67)
Thus we have narrowed down the region of our Ignorance but not
completely
It mll be noted that the two sntuatlons have not been treated
symmmxutally A always,berm; the reliera-m.mm. It might well
happen that X lay made the budget hne of the pomt X Assum
mg for the moment that knowledge of the complete preference eld
were avatlable for what values of X" would tlns be possnble? Ap
plying the requu'ement that X must cost less than X" at the prices
Pb the boundary of this region Wlllbc gwen by the equahty
z PX= z Mb (68)
Geometrically, this locus is the result of swinging the budget line
through the point X , and determining for each position the point
where it reaches the highest indifference curve, to which it is
clearly tangent. If we connect all such points, we have a familiar
oercurve. All points above this offer curve, are denitely better
than X , in the same sense that all points under the budget line
of X are denitely inferior to X. (It is to be emphasized that a
knowledge of two points does not give us this locus in the way that
it gives all the previous boundaries. Nevertheless, given two such
points, it would always be possible to decide just where X" was
with reSpect to X .)
We have now narrowed our ignorance down still further. Ac
tually this is as far as we can go on the basis of the given data.
Note that the old boundaries of our ignorance derived from the
utilization of quantity data alone have been made obsolete by the
additional light furnished by the auxiliary price information.
We have gone a long way, but there is still left a denite area
of darknessthe space between the two heavy lines. I should
like to state as strongly as possible that this nal indeterminacy is
intrinsic and inherent. No amount of ingenuity can remove it,
grounded as it is in the fundamental convexity properties of the
indifference eld, or more accurately in the consistency behavior of
the individual. It is important to prove this rigorously, for pe
culiarlrin the literature of index numbers is an attempt made to
search for limits within which the truth must lie without at the
same time investigating whether or not these are the best possible
limits. Moreover, the limits themselves are sometimes derived
under special approximations, such as the neglect of squares of
small quantities, etc.
To see that these are indeed the best possible limits under the
circumstances, let us suppose that someone proposes narrower
limits. Since our preference eld is arbitrary except possibly for
certain curvature properties, we can draw the true indifference
curve through X so as to contradict any more denite result. if
the proposer claims that a given point in the area of darkness IS
worse than X , we can pass the indifference curve below that point
but above the correct lower boundary in such a way as to make this
statement wrong. Similarly, the opposite statement can be shown
not to hold universally. (It is of course understood that we do
150 FOUNDATIONS OF ECONOMIC ANAL you
not actually change the Indifferenceeld of the economlc umt under
observatxon, but we" can nd a consxstent eld for whtch the
gwen result holds ) Where the offer curves bend back we are able
from quantity consrderatnonsalone to extend our boundanes along
the dotted lunes lndxcated 1n the gure
For well behaved Indifference elds wrth the proper contmuous
denvatlves, the offer curve whlch gwes the upper boundary of the
area of darkness Willbe tangentral to the mdnerence curve gomg
through X and to the lower budget boundary through X Even
If there xsa corner the Indifference curve at X ". there must be
generalnzed tangency the sense of touchmg from above wnthout
crossmg One mlght be tempted to say that the area of mde
termmacy narrows down to a pomt m the vxcmlty of X, but
nothmg IS gamed thereby Nevertheless, the mathemattcran :s
tempted to conSIder Taylors expansnons ln the neighborhood of
the X pomt and to neglect terms of hlgher order Perhaps there
IS even some emplncal stat1st1cal usefulness 1n the handllng of
budgetary data to be found m th1s practlce, Wthh has been assoc1
ated With the names of Bowley and Wald
But from a fundamental vrewpomt there lS no gettmg around
the fact that for any mte move, howeversmall, there remams a
regzon of 1gnorance, th1s reglon may narrow down as the srze of
the movement decreases, but xt never vanishes, except for the
tnvral se of a vamshmg movement Thus. there ISnothing t
be gamed for the present purpose from followmg the practtoe of
DlVISla" m workmg thh dnerentlals or rnmtesrmals The fact
that these avo1d dlcultles connected wrth the tlme reversal and
factoral reversal ISnot an 1nd1catlonof thexr superiority, but of the
fact that they srdestep the mtnnsxc dxcultles of the subject
matter
By the exammatxon of value sums we can somet1mes, but not
always, state denitely whether one sxtuatlon ls worse than another
Butwe can never by these means State 't'na't two sttuatlons are
equally desnrable In fact, as wnll become evxdent from later dlS
It may be worth mentioning that even a knowledge of the curvature of the m
da'erencecurve at X wallnot enable us to narrow down our reglon ol Ignorance It
mll enable us to assert wrth condence that the mdn'erencecurve wall approach arbt
tranly near to the escalating cucle, but the departures for any mte small movement
may be of either 5130and of any magmtude
"' F Dwzsra, EconormqueRatwmllc (Pans 1928)
TRANSFORM'A TIONS, COMJIIODITIES, RA TIONING I 51
cussion, knowledge of prices and quantities on a nite number of
points will not permit us to determine the equality of two points;
but in the limit as the number of points becomes innite we can in
fortunate circumstances determine points of indifference.
This result may be approached by a consideration of the addi
tional information which a third point will yield. We now have
three pairs of points, and it may turn out that any two of them
when put to the test of our computed value sums gives a denite
answer. In that case each pair can be considered by itself without
regard to the third point. Of course, from the discussion of chapter
v it will be clear that a consistent ordinal preference eld can never
give contradictory testimony whereby point A is better than B,
which is better than C, and at the same time the latter is better
than A. For the relations of ordinal utility are transitive.
But the relations of better or worse as revealed by the value
sums are not transitive; that is why I have elsewhereproposed a new
notation to represent revealed" preference in this special sense.
Thus, if our value sums give a denite result so that
2 PX }: PX, (69)
this fact may be represented by the symbol
X " @ X . (70)
Since
X" @ X implies U(X") < U(X), (71)
and this in turn implies
U(X) <1:U(X), (72)
then to avoid a contradiction in logic, we must be able to state
the theorem
X"@X implies X@X".
But this is not the same thing as the meaninglessassertion that
Xb @ Xl implies X X", (73)
or the incorrect assertion that
X @ Xl and X @ X" implies X @ X. (74)
The most that can be stated under the above hypothesis is that
X @ X. (75)
Thls IS much weaker than transrttvrty An additional Indicatxon
that the algebra of revealed preference ts quite drstmct from that
of cardinal or ordmal numbers 15 the fact that equahty 15not
dened consequently, two pomts cannot be placed m one of the
three categories A worse than B, or B nor-se than A. or the tno
equally good All we can say IS either A :s revealed to be worse
than B or B 15rexcaled to be norse than A or there ISno Indlca
tron one way or another These are mutually exclusne categones
only If a consrstent preference eld ls postulated
It rs precxselybecause of this lack of transnthty that knowledge
of a third pomt may add knowledge to a companson bet een two
gwen potnts Our value sums may gne no Inchcatton wzth respect
to pomts X and X taken by themselwes but an Intermedrate pomt
X may serve to Indicate their true ordinal relattonshlp to each
other We are m a posmon now to mchcate Just how much pomts
X and X may narrow down our Ignorance relationshlp to X
If the two addlttonal pomts both he m our old area of darkness
ne are no better off than before But nfX hes m the upper regron
of certamty w1th respect to X and X hes 111the upper region of
certamty wrth respect to X. then even though X hes m the X' s
reglon of uncertainty It can stnll be demtely sand to be better
than X Therefore we have narrowed down our area of darkness
How much can thrs area be reduced the most favorable case?
From the geometry of the problem It can be shown that )1"5must
he on X s o'er curve for the best results We then proceed to
draw the offer curve through X Tlns wnllcross the old one and
thereby narrow our reg1on of Ignorance since every poxnt aboxc
the new offer curve 15denitely better than X If now ne let A
take each and exery posxtronon the old oer curve the process can
be duphcated as many tlmes as we msh, gn mg us a one parameter
famlly of new offer curves The lower envelope to thns family of
curves gwes us our new upper boundary of uncertamty
In the same way, ne can secure a best lower boundary by lettmg
X" travel along the ongmal budget lme, generating through each
such pomt a new budget lme or a one parameter famxly of such
lunesm all The upper envelope of ats farmly of lmes ts our new
lower boundary Our new boundanes must of necessrty he wrthm
the old ones but there ISstlll left an area of darkness
If a fourth pornt lS added we may narrow our Ignorance stnll
TRANSFORM'ATIONS, COAIMODITIES. RATIONING I 53
further; similarly with a fth point. It is intuitively clear that
the boundaries will never meet for any nite number 0f POntS:bUt
in the limit as the number of points becomes innite, the upper and
lower boundaries approach a common limit, which is of course the
indifference curve through X . Points along this curve, and only
such points, can never nd their place in a nite chain of points
which serve to relate them unambiguously to the initial situation.
Thus, with a single dimensional innity of points any indifference
curve may be traced out. With a two-dimensional innity of
points, the whole indifference eld can be determined. In fact,
with this much knowledge we may dispense completely with re
vealed preference, and instead integrate completely our elements
of slope at each point into a one parameter family of indifference
curves.
The above discussion is concerned with the narrowing of our
eld of ignorance under the best circumstances. in actual practice
we shall not necessarily do nearly so well. An instance is provided
by the frequently met case in which we know a complete budget
line (expenditure path, income consumption curve, etc.) giving the
behavior of changes in all of the commodities with changes in
money income but with unchanged prices. Such a situation is
provided by observations of the behavior of more or less similar
people, all confronted with the same prices but differing with
respect to total expenditure.
Not only is this of considerable statistical importance, but it is
of special relevance in connection with the usual economic theory
of index numbers, where a comparison is made between two price
situations, rather than between two particular price-commodity
situations. By applying our previous analysis it will be seen that
the knowledge of an innite number of points along two budget
lines will not be sufcient to determine the whole preference eld,
or even a single indifference curve, or even be sufcient to enable
us to match points of equivalent satisfaction on the two budget
lines. If we select a given point on one line and through it draw
a budget line and an offer curve, then these last two loci will split
the other budget line into three parts. The lower part will consist
of points all worse than the initial point, the upper part will consist
of points all better than that point, while the intermediate part will
constitute our area of indeterminacy.
154 lUUlVUJJUdUF thztvtulu nzvnuluzo
Of course the closer together are the two prxce situatxons the
smaller w11!be the mdetermmacy Therefore we can narrow down
our Ignorance by a knowledge of many Intermediate budget lines
In the hunt as the number of budget lmes becomes Innite such
a way as to decrease 1nden1tely the distance between each we
shall approach the complete Indifference eld
The value sums conSIdered up lll'ltll now have all had the d1s
advantage of bemg capable of yleldmg mdetermmate results We
shall put up wath this If need be but rst we must dctermme
whether xtISnot possrble to devnsean Index of quanttty computable
only from the prxces and amounts of all the commodities which
W111 be an unfathng 1nd1catorof ordlml utthty Can we not nd a
magic formula wh1ch wnll have these properties when apphed to
any preference eld whatsoever or at least to those of the proper
convemty>
The answer 15no As yet no one has dewsed such a formula
and the followmg mathemat1cal reasomng shows why no one exor
can even xr we permit more general functlons than snmple value
sums Any such general formula wrll from the nature of the
problem be a functxon of quantltres and pnces m one or more
(usually two) s1tuatxons Without loss of generahty we may con
srder the puces and quantltzes of all but one of the Situations as
xed whlle the one sxtuatlon represents marbltrary varnble poxnt
In terms of ltS pnces and quantltres (P A) we must be able to
construct a smgle Valued function Q which IS constant so long
as the X pomts remaln on the same mdl'ercncc cune of the
(unknown) preference eld Because this xsto be a quantxty 1ndex
xtxsclear that relatu e prices only can be Important and so we may
express all prlces m terms of the rst good as numerau'e Then Q
may be wrxtten as
= 6 ,, - 76
Q (at; x P:
P1 pa
P1 ) ( )
For observable equnhbrlum quantntles the ratlos of pnces are equal
to marginal rates of substltutnon (U?2 1R") or to ratros of
margmal utxhtles (U2/ U; U../ U;) Although the form of the
relationship wnll change from preference eld to preference eld
and IS m any case unknown to us there W1llstlll nevertheless ln
h specxc mstance be a functlonal relatxonshlp between these
ratios and quantities of goods. Therefore Q will be a uniquely
determined function of the x5 alone. In fact, since Q 13constant
along an indifference curve, it must itself be one cardinal index of
utility; its partial derivatives with respect to each commodity
(computed by adding to the direct effect of each x on Q all indirect
effectsvia inuence on price changes) can be interpreted as marginal
utilities, whose ratios are equal to observable price ratios, etc.
In consequence of this last fact, equation (76) can be written
in the form
a_Q_ _
9(x11'. ,x", axl/x21"'yxl __)_
xn Q= 0. (77)
This identity is a rst order partial differential equation which the
preference eld must satisfy, and only a limited small subset of all
preference elds will do so. Consequently we have proved the
impossibility of nding a magic formula to serve as a quantity indi
cator in the general case of a consistent ordinal preference eld.
In actual fact it is customary to restrict the form of a quantity
index still further, to require that it be a homogeneous function of
the rst order in the component quantities, etc. This restricts the
range of applicability still further, to preference elds possessing
expenditure proportionality, etc.
For any particular 0 it may be a problem of some mathematical
difculty to determine the exact restriction on the preference eld.
But where 0 represents the Laspeyre quantity index, thought of as
a function of the quantity components of the second point, with
base point xed, it will be immediately seen that the only possible
preference eld for which it is exact is that of a family of parallel
straight line indifference curves. Of course, this is almost a re
ducto ad absurdum, since convexity is denied, and since all relative
prices (of goods which are bought) cannot vary.
The ideal index number is one of the most popular index
numbers. It is the geometric mean between Laspeyres and
Paasches index numbers. A. Konus and S. Buscheguennce,23as
well as S. Alexander (in an unpublished Harvard paper), have
shown that it is exact only for certain hyperbolic indifference
curves. This theorem may be derived by solving explicitly the
" See the reference in H. Schultz, A Misunderstanding in Index-Number Theory,"
Econamcm'ca, VII (1939), 8.
156 FOUNDATIONS OF ECONOMIC ANALLSIS
unphed dn'erentnalequatxon Similar results can be derrved for
other approxxmate" formulae Inch have been suggested
As a result of the analysrs of this section the conclusmn may
be ventured that the Important economic content of Index number
theory resrdes m the fact that xt attempts to utlhzc hunted pracc
and quantlty data to mfcr ordmal preference compansons The
aboxe formulatnon seems best desxgnedto reveal thts essentnl con
tent, and to show the mtnnsxc hmltatxons necessarlly lmolved
In the nest sectxon the same analysts rs shown to bc useful m
connection thh the more commonly met formulation of the Index
number problem
PRESENT Fommmnows or INDE'C NUMBERS
We are now m a posmon to apply our tools of analysns to the
more famlhar branch of Index numbers First, e consrder pnce
of hvmg comparxsons bet een tuo dx'erentpnce srtuatxons The
price of Iwmg ldCCnumber ln gomg from the srtuatlon (X') to
(X) rs famzharly dened as the ratno of the cost of the cheapest
bundle of goods at the pnccs of the second sntuatxon nhnch mll
yneldsatrslactron equnalent to that of the xmtzalSituation to the
cost of the zmtral bundle at the mltxal pnces
We assume as known (say from empmcal studies on Identical
xndrwdualsthh \ arymg mcomes) the pace expansron paths untuo
pnce sxtuatxons (P) and (Pb) These are dened by the followmg
sets of parametnc equations
a. = 3.(P1'. , p., I), (t 1, u) (78)
x : "*(Plbo vpast I) ( I: a n) (79)
Consrder an :mtxal srtuatron (X) on the Imtral expansron path
The 1nd1fferencelocus correspondmg to thrs pomt mil xntersect
the second expansron path m a p01nt called (X) such that
mx) = U(X) (80)
An ldCinumber of the prxce of In mg ISdened as follows
_ ): FX
105_ ___
PX (81)
Slmrlarly,
a be
pa = ___Z
PX
Z PX '
(82)
where
U(X") = U"), (83)
and (X) lies on the initial expansion path.
These respective index numbers are not rigidly related; in
general they are not reciprocals. Since we do not know the hypo
thetical preference eld, (Xb) is not known, and so our index
number cannot be computed.
However, from the considerations indicated above we know
that all of the points on the expansion path (b) fall into three
classes with respect to (X):
1. (X) @ (X) z PaX g z P0X (34)
2. (X) @ (X) z PbXa g z PbX (35)
3. (X) @ (sz) z PX > z PaXa (86)
(X) @ (X) ): PbXa > z PbX.
In particular, consider the equality signs in (84) and (85) and we
get the boundary points dened by the intersection of
2 PK = {: PbX (87)
and
x,- = hi(p1b, ' ', Pub, I). (Z = 1, ' ' ', n) (88)
Call this point (X).

Consider the intersection of
Z PX = Z PX" (89)
and
x,- = h,-(plb, - - -, {9,3, I). (i = 1, - -, n) (90)
Call it CX). By denition

(lXab) (Xa) @ (uXab)_ (91)
Hence,
UOX) < U(X) < U("X"), (92)
0r U(X") < U(X) < U(X). (93)

Since along an expansion path the cost of a batch of goods and its
utility" are monotonically related, we have
2 Pb zXab< Z PbXab < Z Pb uXab. (94)
Hence dmdmg by the appropriate factor (af 1t does not vanish)
we have
W
2 Pb 1X06
W <W'
E PbXab z Pb uAab
<95)
or
Pb (Xab Pb uXab
% < 1< %)? (96)
These are vahd double hunts for our mdc'c number computable
under our specnedassumptzons More directly from the den:
tlon of (X) we have
z PbX= 2 Pw (97)
Hence our double hmzts are
%% < In< %; (93)

By symmetry e haxe
Egg < [< %. (99)

where (01) IS symmetrically dened mth respect to CX)
Ag'un these four 11m1tsare not ngldly related no palr bemg neces
sanly reprrocals Furthermore 1t lollons from the analysts of
the earher seetxons that these 1re the best poss:ble double [units
under our hypothess and ml] never converge to equahty for
comet 1nd1fferencecurves
It ISclear that the tuo upper limits are determmable from pnce
and quantity gures alone The pomts CX) and CX) hon ever
can be computed only If certain subsections of the respecttve ex
pansxonpaths are empmcally known Dropping thlS assumptlon
we may ask the question as to .hat lower hmrts can be computed
from price and quantum!gures alone Tlns bemg a more dlicult
problem the answer wnll1ngeneral be less satlsfactory
Recall that CX) was the 1ntersect1onof the budget plane of
(X) thh the expansnonpath of the second price Sltuatton Smce
we do not know the expansron path we cannot compute ('X')
But nf e restnct ourselves to posntne quantrtxes xt 15always
possnbleto nd a pomt on the budget plane whose cost ls less than
('X) and whxchmust afortwn be a lower [mnt Such a pomt IS
TRANSFORMA TIONS. COMMODITIES, RATIONING 159
that point on the budget plane whose cost at the prices (P)
furnishes a minimum with respect to all points on the budget plane.
Let us designate this point (*X). It may be derived by
mm1m121ng
F = 2:312;be (100)
subject to
E::pc; = %:pci". (101)

This is a constrained minimum problem, and, since the equations
are linear, the minimum is a restricted or boundary one, derived
from the fact that no negative quantities are admissible. Ac
tually, it is easily veried that the minimum quantity (*X) is
(O, O, - - -, rm, - - -, 0, 0) where xis the good whose price ratio
between the two situations is lowest, i.e., ,
b _b
%; s %, (102)
(*X) is easily computed since
*xmab= E PX (103)
pmu
Hence this lower limit developed by Mr. Lerner is as follows:

we:
Z PaXa
=223.233: _ J:
pmuz PaXd p a

By symmetry, [n/p." is a lower limit for I ,where
z- . (105)
Pl" : Pi
In particular cases other lower limits are possible. It is possible
that our actually observed situation (X") is such that
(Xb) @ (X), (106)
i.e.,
Z PXb 2 FX. (107)
Recalling the denition of CX), obviously
z PaXb : Pa lXab
160 I~OUNDATIONS OF ECONOMIC ANALYSIS
and (108)
XFX" < 21 lch,
and so m th1s case we hate :1lower hrmt as follows
Pb b
% < I" (109)
l\ote that thxs holds only 1f (106) IS realized On the other hand
gwen
(X) (X). (110)
we have P X
% <I (Ill)
Except m the comcxdental case where the equality Sign holds these
lnmts w1llbe worse than those deduc1ble from the hole expansmn
path They may be better or worse than the Lerner hmlts gwen
m (104) dependmg upon the case selected Note that 1t 15Im
possxble for
(X) (X)
Slmultaneously With
(X) @ (X)
Hence 1t 15 not possnble to derwe lower hnuts to both Indexes
Simultaneously In fact given
(X) @ (X)
and
(X) @ (X).
1t ls 1mpossnbleto compute elther lower lmnt thus manner
If we wndenour assumptlons as to mxtlal Infornlatxon stlll other
hmtts are possxble Knowledge of a thnrd pomt may be utlhzed
by the methods of the prevnous sectton, as may also knowledge of
any 1ntermed1ateexpansxonpaths In fact m the hrmt as all mter
medlate expansxon paths are known 1e as we know the functions
xi = h (P! a Pu I): (' = 1! ! n)
the 1nd1erencemap Itself may be solved for nmphcntly
It IS posmble to develop sxmllar relatxonshnps between usual
quantlty 1ndexes Let us dene our quantity mdex from the ath
I do not d scuss the effect of xmposmgthe assumption of monotomc expanswn
sance there xsno reason to rule out mfenor goods
JKANSFURMA TIONS, COMMODITIES, RATIONING I61
situation to the bth as follows:
Qa=ELJLZP
21wa
):?wi n
=F1? (2)
i.e., this index is the ratio between the actual cost of the batch
()F)to the cost of the cheapest bundle at (Pb) prices which would
yield an equivalent satisfaction to that of (X).
Similarly,
a }: PX I
Q"= W =FF (113)
Obviously, if one situation is preferred to the other, the index
number between them will be less than one, i.e..
U(X) < U(X") implies Q < 1, (114)
and conversely. Likewise
U(X) < U(X") implies Q > 1. (115)
Hence,
Q > 1 implies Q < 1, (116)
and
Q > 1 implies Q < 1. (117)
More generally,
Q = 1 implies Q = 1, (118)
where the inequalities are to be taken in the indicated order.
However,
1
Qab 75 ?; ! (119)
except under special circumstances. The above relations are
consequences of the concavity of the indi'erence loci.
Of course in the absence of knowledge of the preference eld,
it is impossible to compute these index numbers, since we do not
know (Xb)and (X). From the last section we know that
2 P xx :. >: Pb lx < }: PX< 2 Pbax. (120)
Hence,
PbXb PX" Z PX
%FFFF>Q>TFFF' (
or PUP ): P aXs E PX
?! = :: ! xb
> Q > zmr- (122)
As before. ('X) can be computed only If c know the cxpansmn
paths From (103) we nd
W
ZPX '
KW
p.: EPX (123)
Therefore.
p. }:PX'
Pa. E Pch "=- zXFX
Pblx > O> _);PX
2 P6X. n Paasche (124)
Smularly,
EPX> EPX >QM>PX_ 1

Pt . PX .=. E 1 ' P L.15m)
res"
As before. 1! we haxe
(X') @ (X).
0!
(X) (X).
still other lIrmts an. poss1blc
Now 1t 15only m the case here the Paasche Index rs greater
than one or the Laqu res" Index rs less than one that It ISp0551b1c
to say u hrch of the srtuatlons ISbetter But, and thzs asv.hat makes
thls whole branch of theory a stenle excretse, It ISalw'tys possrbleto
determine thls hct v.:thout any hmxtsor \\ ttllout et en Introduczng
such Index numbers Ior our question 13nlwus ans cred, hen
nt can be answered by the prevnous amlysxs of rex caled" prefer
ence, and the Introductlon of quantity 1nd pnce Indexes asInd1rect.
unnecessary, and misleading It as 1ndzrect and unnecessary be
cause lt :s a deduction from the prex 10115Simpler analysw It ts
mlradmg hem nf the tendency to mm:11
surmmnreto the
numerical value of the Index computed There IS not a smgle
vahd general theorem m the present eld of Index numbers wh1ch
13not deduenble from the analyszs of the prevrous section
Of course, 111the case of expenduurc proportronaluy certarn new
unvarxances do emerge (1) 1t nsahmys possublc to dcr-we double
" See the penetrating remarks of W Leontlcf Composxtc Commodtt cs and the
urohlern of Index Number: Economemm vol IV (1936)
TRANSFORJMATIONS, COIM'ODITIES, RA TIONING I63
limits; (2) The respective index numbers are reciprocals, i.e.,
QbaQab = 1 = IabIba_ (125)
One last serious drawback to the present formulations of index
number theory rests in the fact that these are expressed as ratios.
If we admit productive factors as negative commodities, and for
many purposes this greatly enlarges the generality of our analysis,
denominators may vanish or change sign. Since all we wish is an
algebraic comparison between value sums, it is unnecessary and
undesirable to work with ratios. Instead the methods of the
previous section are preferable.
PURE THEORY OF CHOICE UNDER RATIONING
Thus far in this chapter we have discussed the eectsof general
transformations of our dependent and independent variables.
This led naturally into a study of composite commodities and of
index numbers. At this point there is also suggested the interesting
and important related problem of placing constraints upon the
consumer in addition to that imposed by xed total income. Of
course, rationing involves such constraints.
The simplest kind of rationing is that in which the government
speciesthe maximum amount of a particular commodity that each
individual can consume. Sugar provides a common example, being
relatively homogeneous and universally in demand. Ordinarily in
rationing the individual is subjected to an inequality. He must
not consume more than a given amount, but he need not consume
as much as that amount. Of course, if the rationing is to be of
any use, it will be applied to situations in which for many indi
viduals the allotted amount will actually be bought. Otherwise
the law has no teeth, and individual behavior is left unchanged.
The single individual maximizes ordinal utility as before except
that he is now subjected to additional constraints of the form
x; 1);, x,- bj, ' ' ', (126)
where the commodities i, j, - - - are given maximum individual
quotas of respective amounts per unit time, bg,b,-,- - -. To present
the conditions of equilibrium in these circumstances we must dis
tinguish carefully between various possible cases. The Simplest 15
that in which the rationing quotas are so small that each 15effective.
in the sense of expenditure on future goods? Upon reection I
think the reader will conclude that this is primarily a verbal prob
lem, much aggravated by the customary looseand ambiguous usage
of the concepts of money" and numeraire.
Various conventions are open to us, any of which is satisfactory
provided that it is adhered to consistently. We may include
among our commodities goods of dierent periods of time, and we
may assume certain expectations with respect to future prices.
For the present purpose it is simpler to assume that total expendi
ture upon present commodities can differ from present income by
the algebraic amount of saving without going into the problem of
the form (cash, securities, etc.) that these savings take. If saving
is determined by the usual preference calculus, certain implicit as
sumptions are made with respect to future incomes, prices, ordinal
rates of time preference, etc.
If it wishes to, a powerful government can arbitrarily limit the
consumption of all present goods, and at the same time permit only
an arbitrarily small amount of the surplus of income to be carried
over for future uses. Ordinarily, in connection with a rationing
program the State does not so limit saving, relying instead upon
a personal tax program to bring about this desired result. Further
discussion of this point may therefore be deferred.
Single commodity rationing has certain shortcomings which
give rise to a demand for some form of point-rationing." Instead
of limiting the amount of a single commodity, the individual is
limited to a weighted sum of a number of commodities, the point
prices providing the relative weights. But such a weighted sum
represents nothing more than a single commodity in a new trans
formed set of variables. It is for this reason that the general
theory of rationing belongs in the present chapter on transforma
tions and composite commodities.
It would lead us into the eld of welfare economics if we were
to discuss the criteria used in classifying commodities into groups
and the determination of their point values. Sufceit to say that
administrative considerations, criteria of substitutability on the
consumption and production sides, all contribute to the decision
as to which commodities shall fall in the same group, what the
relative values of each shall be, how many groups there shall be, etc.
Moreover, although this is not strictly necessary, each comrr1 v
166 FOUNDATIONS 01' ECOOMIC ANALLSIS
rs usually restncted to but one group so as to obvnte the necessnty
of handlmg many kinds of tokens or stamps m makmg a gwen
purchase
In addition then to the dollar prices (12;. . p),and money
Income. I, each consumer IS confronted thh r classes ol pomt
DFICCS(pl'n ! pu,)r (Pin. 1 Pu"): I (pl" | Pu'), and led
totals of expendable pomts per umt time (I ', I ', , I') Each
ol the sets of pomt pnces mll lune zeros {ormost commodtttcs and
for a gwen commodlty all but one pomt pncc wzll usually be zero
Thxs bemg the case, the consumer \ull mammlze U(x., , x..)
subject to the generalized budgetary constraints
? [9,345,
g I: ? Px'xi E I" c }; Pfx: 5--I' (128)
For the moment the matnx [pf] wallbe assumed to be of rank

(r+1) There mll result optimal amounts of each good for each
specied complete set of pomt prices and e\pendlture allotment
We may summanze tlus result by \\ ntmg down the generaltzed
demand schedule for each good, wntten as a functlon of all pnces
and pomts and all total mcomesor expendttures Thus
x: ==h'(P1 Pn PI, upal. ,Pl'. rpn'o Il I'. t It)
(1 = I , n) (129)
lt 15the purpose of the regulating authorities so to determme pomt

pnces and allotments as to result m equ1t'1ble" mter mduwdual
amounts of consumptlon, appropriate total output at appropriate
dollar prices In addmon to such pohcy matters, the econonnstas
Interested the purely posntne problem of determmmg the prop
erttes of these demand functnons Implled by the ordxnal manmxza
tion process
For this purpose we must examine the maumum conditions 0f
ethbnum By the usual Lagrangean multipher techmque the
rst order conditions are casxly seen to be
U.+M>.+>~p.'+ + :=o, (:=1, ,n) (130)
provnded that all of the group allotments art effect e' as md!
cated by the presence ol the equality sngns m (128) If any PW
txcular group allotment ISIneffective for the 1ndn Idual m question
1m nor-wt 5m 1 1U.-\.5, 0.11.LUDITI ES. - %TIONING 167
its Lagrangean multiplier may be set equal to zero in the above
equations.
In words: the consumer will purcha each good up to the point
Where its marginal utility is equal to a weighted average of its
various dollar and point prices, the weights being the marginal
utilities of the last dollar or group ration coupons}1
For a regular maximum our necessary and sufcient secondary
conditions are contained in the statement that the Hessian of the
utility function must represent the matrix of a quadratic form
which is negative denite under the (r + 1) linear constraints.
This is equivalent to certain conditions on the matrix formed by
bordering the Hessian with the matrix: of the constraints, the
transpose of the latter, and zeros. If we strike out rows and
columns corresponding to each of (n.
r + 1) commodities in
turn, the resulting (n -r + 1) principal minors must oscillate in
Sign, the smallest one being negative, the next positive, etc.
Thus let
(i)]. = 11"r7")
Am: [U
19,- pik]
O (kt: = 01' "1 f)
[m=(1zr+1),---,n] (131)
Then
( 1)A > O. (132)
As throughout this book, the fundamental observationally

meaningful restrictions on observable prices and quantities emerge
as consequences of these secondary extremum conditions. In this
case such implications can most simply be stated after the concept
of a compensated price change has been suitably generalized. .At
least since the time of Slutsky it has been customary to deal with
a compound change in price and income, where the latter ISvaried
along with the former so as to leave the 1nd1v1dualon the same
indifference curve. ._
When rationng enters the picture so that we have auxiliary
constraints, the problem is made more complex. An mcrease m a
This seems to have been pointed out in print for the rst time by T. Scto);
The Political Economy of Consumers' Rationing," Reviewof conomxctahslllcsi;_
(1942). 114-124. Other theoretical aspects of rationing are discussedm H. Ill (lags-:5.
"Theoretical Aspects of Rationing," Quarterly Journal of Economics, L\ __( ,_
378-397. Also see N. Kalecki, General Rationing," Oxford Bullet of Stan: ms,,
III (1941).
X68 I-UUNDATIONS OI" ECONOMIC ANALYSIS
given price. or in a given point price, will make the individual

worse off. But he can be compensated for this in many different
waysby a change in money income. by a change in his point
allotments of the same or of a different group, or by any combina
tion of these. It is most useful for the present purpose to consider
that special compensated change in the dollar or point price of a
commodity in v.hich the individual is kept just as well off as belore
by means of a simultaneous change in the same total expenditure
allotment (dollar or particular grOUppoint allotment). Let the
change in the 1th good resulting from a change in the kth group
point price of the jth good when compensated as above be written
in the form (x./p,*)p Then regarded as an n by n matrix.
with k a xed number, this must be negative semidenite ol rank
(n r + 1). Thus, the compensated change in a good with re
spect to one of its own point prices must always be negative; in
addition, we have inequalities on certain cross product and inter
action terms These are expressed completely by the statement
that the smallest naturally ordered principal minor of the above
matrix must be negative, the next positive, etc., and nally the
(n - r + 2)th and all higher principal minors must vanish. The
last part of this statement is equivalent to the obvious proposition
that the demand functions are homogeneous ol' order zero in each
set of point prices and expenditure
If one wishes. off diagonal terms may be interpreted as general
ized complementarity coefcients, and it may be left as an exercise
to the reader to show that these obey the usual symmetry rules
In fact. the reader may easily show that the matrix of compensa
tion terms is nothing more than the northwest partition of the
inverse of the A. matrix, after this submatrix has been premultiplied
by the scalar ( X)
The interested reader may also develop for himself the general.
ized Le Chatelier-Braun principle introduced in chapter iii. ln
words, its economic signicance may be summarized as follows: if
in a given position of equilibrium a (compensated) change in price
is made, the resulting change in amount demanded of that good
will be greater il the individual is not subjected to the extra con
straints of rationing than if he is subjected to such constraints:
A
, " See chap v, p 103 Also see p 378.
TRANSFORMA TIONS, COMMODITIES, RATIONING 169
furthermore, the introduction of each new constraint will make
demand still more inelastic.
The tendency for demand to become inelastic under rationing
has a number of important implications for policy. In particular,
it helps to explain why in a system which has already been subjected
to numerous direct controls, scal measures aimed at sopping up
superuous income have effects which are of only secondary impor
tance. To explore this further would involve us in considerations
of effective demand outside the scope of the present discussion.
Most economists will not be interested in the somewhat arti
cial compensated change in point-price, but rather in the ordinary
ceten'sparibas change, in which other point-prices and expenditure
are kept constant. It is easy to show by explicit differentiation
of the equilibrium conditions, as well as from general considerations,
that this may be split up into two parts: the compensated change
and the income effect. Thus,
(axi/apjk)t. par.= (axi/apik)comp._ xi(axi/aIk) (133)
Except for (generalized) inferior consumption goods, the second
term will be positive; consequently, a change in a good resulting
from an increase in its point price must be negative'in all normal
cases. But if we have generalized inferiority, we may (but need
not) have a generalized Giens paradox, in which raising the
point-price of a commodity causes more of it to be bought.
Our discussion of rationing may be brought to a close with a
few comments on the special mathematical features which dis
tinguish it from the general analytic case of any linear constraints.
First, it is almost universally the custom to make all point prices
positive, even though one could in theory imagine a case in which
there were negative prices. (For example, ration tickets might be
given to a consumer who will take certain redundant staples, etc.)
Second, the authorities often set ration point-pricesproportional
to dollar prices. A special case of this is provided by dollar ex
penditure control on various commodities or groups of commodi
ties. An even more special case is that of over-all expenditure
control, in which the individual is given a quota of dollar expendi
ture, ordinarily less in amount than his disposable income. If
savings are not treated as one of the n commodities, this is equiva
lent to introducing linear constraints whose rank is less than
(r + l); consequently the equations of (128) are inconsistentll
equality signs are used We must then have recourse to inequali
tics, of course, in this case the dollar budget equation must bi
relaxed in favor of the new quota allotment. In rare casesiiheri
ration tickets or quotas are over-generous, the auxiliary constrain
will become an inequality and the rationing will be ineffective
A different mathematical problem arises if the consumer ca:
pay for a commodity by giving up a certain number of ratio:
stamps of one kind or by giving up some of another. Depending
upon the relatixe prices and scarcities of the different stamps, ill
will ordinarily choose to spend one rather than the other in :
fashion reminiscent of Gresham's Law. If he has similar option
with respect to many commodities but at different relatiie ratios
he will spend a given kind of stamp on those commodities in whicl
it is relatively advantageous in a manner formally identical wit]
the classical theory of comparative advantage as applied to th
determination of which of many commodities will be exported am
which Will be imported.
The above case merges into that in which the different kind
of stamps can be converted into each other at rates determined 0i
3 black or white market, or by the government. Unless th
government were explicitly to ban such transactions, there ouh
inevitably arise trade in the different ration groups and money
Arbitrage would create ruling market ratios of exchange which an;
indiVidual as a small consumer could not appreciably affect. Ont
the indiVidual is confronted with given rates at which he can bu;
or sell each type of ration points, the problem is no longer one C
Since In Practice a commodity is not given more than one ration pol! Pno
degeneracy" cannot arise In any other nay. If the authorities should require multh
pouit prices for a given commodity, unless mre is taken the consumer may be l"ut "
the posmon of not being able to spend all his pomts This difficulty may arise unt
dierent ways through tmc degeneracy and inconSistency in the auiiiliary coristl'a1t
or frOmthe fact that the admlSSlblesolutions of the linear equations do not yield Po
amounts of all commodities When this is the case. the rst order condIUOSOfequ
Iibrium given in (130) are modied, but not as before by setting certain of the LigizlgI
multipliers equal to zero. but rather by replacmg certain equality signs by greater tha
signs Any commodity which is not bought Willhave a marginal utility les: than it
above speCiedWeighted average of paint prices The problems raised ln this [comm
are Similar to those discussed by Schlesmger, Viaid, v Neuman, Nasser, and v brackeleI
_m connection with the conSistency and independence of the equations Of the Walmsu
iry of production in its Simplest constant coecxent form
44\aa..|.vut VA\JILJALAV4VU
UVJVJJIJUUJIJO, nllUlVllVU I71
many auxiliary constraints. Thus, if orange stamps can be bought

or sold for ve cents apiece, and if a commodity costs twenty-ve
cents and two orange stamps, it can simply be regarded as a com
modity which costs thirty-ve cents. The consumer simply reverts
back to the standard type of budget equation, modied only by
the fact that his money income may be increased or decreased by
the need to buy or sell ration points; or, what is simply another
way of saying the same thing, the individual has higher incomeand
is confronted by a new higher set of prices. If the dollar values
of the rst, second, - - -, and rth ration points are (b, b, - - -, b'),
the new budget equation becomes
% (p,-+ )Ebkp,")x,-
= I + gap. (134)
Actually, from a welfare point of view it can be shown that the
free interchange of different kinds of stamps, among themselves
and against money, is in a certain sense optimal. In fact, it is an
indirect way of permitting individuals to exchange goods so long as
the exchange is mutually advantageous. These last two sentences
must be qualied. They must not be taken to mean that every
individual will be better off if coupons or goods can be interchanged.
Thus, if the rich buy redundant ration coupons from the poor to
the mutual advantage of both, their expenditure by the rich will
cause an increase in the point price of the scarce rationed goods in
question to the disadvantage of the middle classes.
Nevertheless, it can be shown that free interchange is optimal
in the sense that its introduction accompanied by appropriate
modications in the point allocations to every individual could
lead to an improvement for everybody. In the above examp'e,
the middle classes could be bribed into acquiescence, and there
would still be a margin of advantage left for rich and poor. I do
not imply by this that the middle classes shouldreceive such a bribe
since that would suggest a belief in the perfection of the previous
status qua. Nor should it be thought that anything said here is an
argument for making coupons interchangeable, since there might
very well be in actual fact very grave difculties in the way of
devising a method of point allocation which would recognize the
harm done to particular individuals.
CHAPTER VII
SOME SPECIAL ASPECTS OF THE THEORY OF

CONSUMER S BEH AtVIOR
CHAPTERv exhausts the content of lltlllty analysns m 1125most

general form mvolvmg only an ordmal preference eld There
remams the hterature a great number of dlscussmns of particular
problems wluch mvolve specml and extra assumptrons In order
to present a fanrly complete account of the present status of the
theory I propose to examine some of these carefully to show their
empmcal meanmg Tlns Involves a break 111the umty of exposn
tron s1nceeach specnal assumptlon has often been made 1ndepend
ently of all others There IS no ch01ce but to go through the lnst
wnth no regard for contmulty Among the toplcs to be discussed
wrll be the cardinal measure of utility mdependencc of utlhtles
and measures of complementarity and constancy of the marginal
utlhty of money
It 15clear that every assumptlon either places restrlctxons upon
our empmcal data or xs meanmgless A price must be pad for
any sunphcattons Introduced mto our bas hypotheses Thts
prlce IS the hmntln;5of the eld of applicablhty and relevance of
the theory because of the extra empmcal restrlctlons to be tm
posed on the data Many wnters do not appear to be aware of
thts, 1nany case few have mdxcated the costhness of their assump
tions or have adduced any evndence to support a presumptxon of
then adm1531b1hty
There IS a further senous dtlculty Desptte the fact that
developments m thus eld are not recent and that mathematical
methods of exposxtron have been employed ambxgutty strll per
meates the contentlons of many writers This amblguxty can go
unnotlced precisely because there has been so httle Interest m the
operational Slgmcanceof these assumptlons To put the matter
somexxhat harshly amblguously dened assumptions are used to
gwe a semblance of derwxng theorems which are themselves
oncluswe
SPECIAL ASPECTS OF CONSUMER TIIEOR Y 173
A recurring source of difculty in this connection which goes
back as far as Marshall is the practice of introducing certain mathe
matical relations as alleged approximations." These are pre
sented as being valid in the neighborhood of a point of equilibrium.
Even if such a relationship be admissible (as in the case of Taylors
expansion), precisely because it can be applied to any and all
properly continuous functions it is devoid of meaningful sig
nicance. Moreover, in common usage the restrictions which are
introduced as allegedly holding only for a restricted neighborhood
of equilibrium are in fact used to deduce results which follow only
from an entirely different interpretation of these assumptions.
We shall have occasion to enter more deeply into these matters at
later points in the discussion.
THE CARDINAL MEASURE OF UTILITY
We have seen from chapter v that all empirical market behavior
is independent of the choice of a particular index of utility, or
indeed of the choice of any measure of utility at all.1 Nevertheless,
many writers have wished to introduce the concept of a cardinal
measure of utility, unique except for constants of scale and origin.
Lange,2 Fisher,3 and others have contended that measurable
utility, while superuous from the standpoint of positivistic be
havioristic description, is necessary for the purpose of a normative
science of welfare economics.
While I cannot consider this problem in detail, it is well to point
out that this is not at all necessary. Assuming that Welfare Eco
nomics involves comparisons between individuals, it is suicient
that explicit welfare judgments be made such that we are able to
relate ordinally all possible combinations of goods and services
consumed by each and every individual. Nothing at all is gained
by the selection of individual cardinal measures of utility.4
Some writers have partially recognized this fact, but still maintain that the avoid
ance of the use of utility is a stunt," an axiomatic experiment by means of which we
make our way more difcult. From the present operational orientation this view is
clearly supercial. _
2Osr Lange, The Determinateness of the Utility Function," Raie-w of Economic
Studies, I (1934), 218-225. .
" Irving Fisher, "A Statistical Method for Measuring Marginal Utility' and Testing
the Justice of a Progressive Income Tax," in EconomicEssays in Honor of John Bates
Clark (New York. 1927). _ . . _
Of course, if one considers generalwelfareas the algebraic sum of indiwdual cardinal
174 mime/arrows or cconozric ANALYSIS
In any case it is clear that there are an innity of ways of select
ing a particular utility index and dening it as the true cardinal
measure of utility. Thus, one might draw a straight line from
the origin through a point representing an arbitrary bundle of
goods and services. The numerical distance of any point on this
line from the origin could serve as an index of utility, and might
be dignied with a unique title
THE ASSUMPTION or INDEPENDENT UTILITIES
A method of selection such as that outlined above is obviously

arbitrary. Others are more subtle, but equally arbitrary. Pro
fessors Fisher and Frisch,6 employing only price and quantity
data, have measured what is alleged to be marginal and total
utilities and how these magnitudes vary with changes in income,
quantities, etc. Although the exact technique differs between
these two writers. the fundamental principle remains the same.
This can be illustrated With reference to a simplied case where
only two consumption goods are involved.
Given a great number of observations on prices. quantities, and
total income. one could in the limit more or less trace out the whole
indifference map We should still, however, havesaid nothing
about the numbering of the one-parameter family of indifference
loci so traced out Both Professors Frisch and Fisher employ the
following denition for selecting out a particular utility index to
be designated the true" measure of utility, subject to origin and
scale constants That utility index, if it exists, is to be selected
which can be written in the form
so= f(x) + g(y); (1)
i e , for which
(?qu _
may = O (2)
utilities, then one WIrequire the cardinal measurability of utility But such an assump
tion is arbitrary and gratuitous Cf A. Burk (Bergson). A Reformulation oi Certain
Aspects of Welfare Economics," Quarterly Journal of Economics,vol LII, no 2, Febru
ary, 1933 pp. 310334
Fisher, up at
R Frlsch. "New Methods of Measuring Marginal Utility," Barrage sur aliena
machen Theone, no 3 (Tubingen, 1932), H Schultz, The Theory and Measurement Of
Demand, pp 111-117
SPECIAL ASPECTS OF CONSUAIER THEORY I75
If there exists one index of utility which can be written in the
form of equation (1), then any other index which obeys the same
law must dier from it only by a linear transformation. For,
consider another index
_ . F = F(so), (3)
for which
62F _ 0
xy : ' (4)
Successively differentiating (3) partially with respect to x and y,
we get
62F I II
a??? = F 99111
+ F (Prior (5)
This, together with (2), requires that
F "90:99E 0, (6)
or
F"(sa) E 0- (7)
Therefore,
F = a + I790, (8)
where a and b are origin and scale constants respectively.
It is clear, therefore, that the assumption that utilities shall be
independent will help to select one utility index as the cardinal
measure of utility. Nevertheless, even this convention is not in
general applicable. It will guarantee us that we do not have two
different utility scales, as has been shown in the above proof; it
will not, in general, provide us with even one scale.
If We assume an indierence eld obeying the ordinary con
cavity restrictions and nothing more, then there will not, in general,
be even one utility index which can be written in the form
90: E 0
Let s write out one legitimate utility index.

H = H(x, y). (10)
Does there exist a transformation F such that
so= F[H(x. y)] = f(x) + g(y)? (11)
I76 FOL/ADATJONS0!" L'CUNUUIC AAAL}SIS
The ansncr m general must be m the ncgame Further arbutrary
restrictxons must rst be placed upon the Indifference eld
Let the mdx'erence eld be dened In the followmg form, mde
pendent ol any utility concept
(Z)n-mm : R(x,y)' (12)

where R ISa function of :: and y obeymg the followmg curvature
requirements
R, - R.,R < 0 (13)
Ihe necessary and sufcunt condition that there cust .: utnhty
Index hlch can be nttcn m the form
sa = f(x) + g(y).
a,,: 0
ISas follous
RR" - R,R., = 0 (14)
or
a! 10;: R
W " (5)
'I he necessuty ls \ernhed by the (h'erent: mon of
R(x >) = (16)
Ihe suPcneney xs also casxb Indicated

If
03log R : 0
xy
logR = log [:(x) log k(y)- log kg-}. (17)
hue h and k are arbltmry functions Form the expressuon
Rd; + d =%x;
dx + d) (18)
[ hrs can easuly be transformed mto the ex tct dlffercntnal
dp = h(x)dx + k(y)dy. (19)
SPECIAL ASPECTS OF CONSUMER THEORY I77
OI
,0= [ha-x + fay/mm + constant

= f (x) + g(y). (20)
We must now investigate the meaning of the restriction in (14).
The assumption of independence of utilities in order to dene a
cardinal measure of utility is seen to involve (1) a convention by
means of which one out of an innity of possible utility scales is
designated as the true cardinal measure of utility; (2) an arbitrary
a. priori restriction upon the preference eld, and hence upon em
pirical price-quantity behavior. The meaning of this restriction
we must now investigate.
The functional restriction (14) is a partial dierential equation
of the second order of the general form
M(Rr R1! RIMRzn Rum R:: x: 3) = 0- (21)
Subject to boundary conditions involving two arbitrary functions,
it will serve to dene a unique solution function
R = (x, y). (22)
More specically, if we are given as empirical observational data
the two expenditure paths corresponding to the changes in quan
tities with income in each of two respective price situations, then
from these observations, and these alone, the whole eld of in
difference curves can be determined by suitable extrapolation.
It is not easy to visualize intuitively why this should be so;
indeed, few economists would be so bold as to claim that the be
havior of an individual in all conceivable circumstances should be
derivable from so few observations. And yet this is the conclusion
to which we are forced by the apparently innocuous assumption of
independence of utilities.
Moreover, (14) places denite restrictions on our demand func
tions. the validity of which are equally dubious and equally im
possibleto comprehend intuitively. For the simple two-commodity
case our conditions of demand equilibrium can be written
;: = R(x, y), (23)
Pv
I=sz+PVy'
These may be transformed mto
x_m
_ PvP"
(.13.: JL)
(24)
y = n (_P 12:
These demand equations must be subject to the restnctxon
2log %
_y:
xy '0 (25)
When there are more than two goods the restncttons nmphed
by the very poss:b111tyof an Independent Index of utlhty take on
a dlerent and more complicated form If there ex15tsan underF
for Wthh
Fu Oo (" # .?) (26)
then
F : F,(P)P) + F(P)Pt: E 0: (27)
where p15any other Index Thus
(at! =
___... 28
M Tu) (a # J) ( )
where T xs an arbxtrary function, and go15 any mdex of utlhty
Takmg mto account the (n 1)(n 2)/2 condltzons of mtegra
blhty th1s xmphes an addlttonal 11(n ~1)/2 condlttons It ISto
be noted that these are ldentmes holdmg everywhere Not only
are they necessary, but the transformahon
F = f e'TWdv (29:
shows them to be sufcnt as well
In terms at the mdtieteme var.-wma.these. take. std]. anothel
form Let
2-,.= .. & __;1
Pr _ dx: __ R(x1 , x,.) (30
These, of course, satxsfy the Identttles
{Ri
r 5 *R (2.1.12 = I. .n) (31
SPECIAL ASPECTS OF CONSUMER THEORY 179
If an independent index of utility is possible, then we must have
a "R5
xk 20. (jki) (32)
In view of the n3 relations of (31), the above n(n 1)2 relations
are not all independent. At most 1z(n2) are independent, and
these may be written in the form
alRi . o I .
xJ. 50' ($95.7)! (?,_7=2,---,n)
lRi (33)
a(12) .:..0_
___1_
6.171
(i = 3, ...,)
These conditions are both necessary and sufcient. They im
ply among other things the (n 1)(n 2)/2 integrability condi
tions of equation (14) of chapter v. On the other hand, if the
latter are postulated at the beginning, then equations (33) cease
to be all independent and can be reduced in number.
We have then n(n -2) partial dierential equations of the rst
order. Subject to an equal number of arbitrary functions, the
general solution is uniquely determined. But empirically an ob
servation of an expenditure path involves (n 1) functions.
Hence, observation of more than n(n 2)/ (n 1), or more than
n expenditure paths, could be used to disprove the possibility of
independence.
In words, the implications of independence are that the amount
bought of any good x,- can be expressed in terms of its price p,,
the price of any other good {J,-,and the amount of expenditure upon
these two goods alone, i.e., upon the amount left over out of total
expenditure after all other goods are bought. Thus, the general
demand function
xi = ]?an ' ' , imI)
can be written in the special forms
(Pn ' ', Pm 1) E hijbi, P131 + Pax.-+ [),-x,- >; kak) (34)
for any i not equal to j. Subject to the conditions of (25) and
certain consistency relationships, these conditions seem to be
sufcient as well as necessary.
4
Except for R. G. D. Allens derivation of equation (15) for the
two-good case. 1 am not aware that these full price-quantity impli
cations of independence have previously been derived. However.
fragmentary sets of necessary but by no means sufcient conditions
for independent goods have been derived by Slutsky and by Milton
Friedman They differ fundamentally from those above, and are
of a species which I choose to call local implications of independ
ence. At a given pomt in the (X) space it may be possible to nd
a transformation F(<p)such that
Fu : FIPU+ F"q0,qa,E 0 (" # ]) (35)
Thus, if there are only two goods (xx) and (x2), this is always
possible at every point so that the local conditions degenerate into
trivial identities At any point (X) = (xi, are),let
F"() _ Mei-Y)
m) mxwxr
and
Flo,(X) = 0. (37)
This differs from the previous non-local conditions in that the

above relation is an eguahty at a pom: and not an identity
In the general case of ?: goods a local transformation of this
type is possible only if the following relations hold,
<Pn(X) %(X) _
991(X)02(X) : $.(X)go,(X) ' ($ 75.7 75 l) (38)
These are [n(n 1)/2 !] independent conditions, and as before
need not be identities. Even if they hold everywhere,independence
is notnecessarily implied. If the local conditions hold, the bordered
determinant of the type given in equation (34) of chapter v takes
on a special form, and consequently certain restrictions upon the
demand functions can be derived.
Slutsky shows that
Pu : (" A (22
P " 0? pmPn+1") ! (39)
R. G D Allen, A Comparison between Different Denitions of Complementaf
and Competitive Goods," Economstnca, II (1934} 168-175
' Milton Friedman, "Professor Plgou's Method for Measuring Elastique: of Demand
from Budgetary Data," Quarterly Journal of Economws. November, 1935. pp 151163
E. Slutsky, Sulla teorla del bilancm del consumatore," Gromale degli Economist!
LI 915), 23-26
where
6x1
K11 K12 ' KI" 6]?
636:
K12 K22 ' ' ' K2n
p = . . . . ' (40)
8 n
Kln K2n ' ' Kan 'ai]:
a] a] 61
and the subscripts indicate appropriate cofactors. Also
a 1
09 = .(I) (41)
in the notation of chapter v.
If an index exists such that
F,-,-E 0, (i # j) (42)
at a point, then
a 1" = ? . (imam) (43)

Pn+1.n+1.12 Pn+1.n+1.ij
These are [11(12 1)/2 1] independent and meaningful restric

tions on the demand functions and completely exhaust the local
implications of the independence assumption.
I should also like to point out that equations (43) might be used
to determine the marginal utility of income if (1) independence is
Possible; (2) that utility index which can be written as an additive
sum of independent utility is dened as the true cardinal index
of utility. I commend it to the attention of the never lacking
army of utility measurers. _
Mr. Friedman has derived the following n(n 1) local inde
pendence conditions: 1
E
"h'i = (Ml _ km,r
m': 1 km,[)+ 2.1[(k5
1m '- kt)1 "'_ kikj(77i'77j)]_
km,r (44)
In the two-dimensional case the restriction is vacuous.
1Friedman, op. cit., p. 162.
From the fact that their number is excessive, it is clear that
they cannot all be independent. In fact, I am unable to prove
that they are complete,i e , equivalent to the above Slutslty condi
tions; indeed, I would venture the conjecture that they are not
It would be literally imposable [or any indiv1dual to determine
by introspection whether or not his demand functions satised the
above relations Indeed, the likelihood that such relations, se
lected arbitrarily from an innity of possible functional relations,
should hold is innitely small The minute amount of plauSibility
" On page 162, footnote 2, Mr Friedman argues against the admissibility of the
Pigou theorem that
k k
Er when =0= (o)
1 "!!! mt '!1!
on the grounds that " there is a presumption that 7mand k; [my notation] are
"it truly related " The whole problem is of course ambiguous until a particular set of
indifference curves are speCied Nevertheless within the realm of probability and
presumption, Mr Fnedmans statement seems to be incorrect In chapter v it i:
shown that
I
Z 12.170
1
=1.
Z k.
I
i e . a weighted average of all income elastiCities equals unity regardless of the number
of goods The average k is given by
Z le.
, = ..l
n n
This approaches zero as the number of goods increases
A survey of empiriml budget studies Will convmoe the reader that the incomi
elastiuties are distributed around unity, while the proportions vary around 1/1
See also A C Pigou and N Georgescu Roegen, Marginal Utility of Money ant
Elastique: of Demand. Quarterly Journal of Economics,vol L. no 3 (1936)
The above Pigou theorem rests, nevertheless upon the restrictive assumption 01
independence If it could be shown that it holds for small values of le./rm in ever? 0355
it would be much more useful
That this cannot be true in general is shown by the followmg speCial case Lei
consUmption of n goods be completely JOInt in the sense that there are always xez
prOportions Then regardless of the number of goods m: = 1. and via/1m = 1 0
the other hand it is easin shown that ii i E np. Consequently. flu/n = ku/h
Even though we make the k : become arbitrarily small as compared to the income 9135
nana, their ratio may take on any value other than unity Thus, the Pigou theoren
() is false
Oi course. this example rests upon discontinuities in the higher derivatives. but l
should thin]. that e could approximate this condition by appropriate chance of con
tinuous derivatives
ornblb AOIDDIO UF LUIVOUJWK
J'HUKY 153
to the proposition that independence is an admissible assumption
comes from the fact that the question is usually posed in such a
way that independence appears to be an intermediate class between
the extreme classes of competitiveness and complementarity. Em
ploying a rudimentary concept of the equal probability of the
unknown, one is inclined at rst blush to acquiesce to the inde
pendence assumption. The error involved in so doing is obvious
from our previous discussion, completely aside from the crucial
ambiguity attaching to the older notions of independence and
complementarity.
COMPLEMENTARITY
In my opinion the problem of complementarity has received
more attention than is merited by its intrinsic importance. Never
theless, as a result of the interest in this subject, crucial incon
sistencies in the thought of Pareto were revealed by Hicks and
Allen, and much light was thrown on the cardinal and ordinal
conceptions of utility.
The older writers, Fisher, Pareto, and Edgeworth, suggested
as a qualitative denition of complementarity between two goods,
9:,-and x,-, the sign of the cross derivative of the utility function;
i.e., goods were complementary, independent, or competitive de
pending upon whether
62 > = . 45
xixj < O ( )
If one assumes only an ordinal preference eld, all numerical utility
indexes are equally admissible. This cross derivative is not in
variant in sign under a change in the index of utility. Consider
a monotonic transformation of a.
dF
U ), d, > O
U;: _piy (46)
dF sz
U;,- = (170 a.-,-+ dPz sa.-soi
It may be argued that regarded purely as a working hypothesis the facts do not
sharply contradict the independence aSSUmption. A little investigation revls that
such a hypothesis has not been tested from this point of view. On the contrary, it is
implicitly assumed from the beginning in the manipulation of the statistical data.
Hence, one would have to go back to examine the original empirical data. It 15interest
ing to note that observations on three expenditure paths would be sufcient to contra
the independence assumption in the case of two commodities.
By a proper selection of the arbitrary term, sz/dlpz, the Slgnof
the new cross derivative can be made to differ from that of the old
As a corollary of this lack of invariance, complementarity so dened
Willhave nothing to do with the budgetary habits of individuals
With respect to the two goods in question. Similarly, an indi
iiduals introspective, spontaneous, intuitive designation of tuo
goods as complementary or competitive (sugar and coffee.beefor
pork, etc ) corresponds not to such a measure as this, but rather
to behavioristic properties of the preference eld and demand
functions
In the course oi the last xe years the world's best economists
have spent conSIderable time and energy in the study of Professor
Hickss Value and Capital It is perhaps symptomatic of the
essential unimportance of the concept of complementarity that in
this period no one seems to have noticed that the author givesmo
or more distinct (and inconsistent?) denitions of complementarity.
The verbal definition in the tent (p 44) seems to be di'erent from
that in the mathematical appendix (paragraph 8)
Hicks and Allen. SlutsLy. and Schultz have suggested invariant. measures ol
complementarity which are properties of the indifference curve system and demand
function Perhaps the Simplest measure of complementarity between two goods 24
and x,. is the sign of
31'. az. K"?!
K = - : = K =
ii ap, + , aI .. or m x;
If there are but two commodities, this must always be posutive in Sign. In the many
commodity case at least one must be of positive Signin order to satisfy the relationships
2. 93K = 0, (. = la | n)
K.. <O
Professor Leontiel has suggested the followmg invariant measure of independence
corresponding most closely to the old notions
.
azdR: , lg( U, )
ax; ' x.:|:, _ 0
Since the above lines ere rst written Professor Hickss two works. Valueam
Capital and Theam: Mathemahqzde La Valeur, have appeared I still believe. however
that Hickss own solid contributiOns to economic theory do not rest on his treatment oi
complementarity, and that the extended discussion of the concept is a tribute to an 01!
love rather than the necessary consequence of the subject matter
J R. Hicks, Value and Capital (Oxford, 1939)
Actually. there is still a third denition in the footnote on page 44 Only I "
case of three goods is it clearly equivalent to one of the other trio. in the general 035"
SPECIAL ASPECTS OF CONSUJIIER THEORY 185
This may be seen in many ways. While the mathematical
denition can be applied to the case where there are only two goods,
the literary denition cannot. (It is irrelevant that in the case of
two goods the mathematical denition of complementarity admits
of only one possible algebraic sense.) There is no reason why two
distinct denitions should give the same answer in any particular
case, so it is not surprising that one can invent examples ad im'ni
tum for which two goods, such as wheat and linen, are on one of
the denitions complements and on the other substitutes. But as
we shall see in a moment, things are in even a worse state. Ac
cording to the denition of the literary text, it is possible for wheat
and linen to be complements and substitutes at the same time,
depending on the selection of the third good which is to serve as
numeraire. The denition is ambiguous; instead of reecting the
properties of two goods, it (or rather they) represents the properties
of three goods.
The mathematical appendix denition gives as a coefcient of
complementarity between the ith and jth good the element of the
ith row and jth column of the matrix ($,-")where
x = #19, U) (47)
represents in matrix notation the demand functions, for each
indifference level.
Because of the homogeneity property discussed in chapter v it
is clear that these n demand functions cannot be inverted to give
each [2 in terms of the x's and U. Only relative prices can be
determined. We can use any one price as numeraire, discard one
of the above equations and invert the others so as to give (n -1)
relative prices in terms of (n 1) goods and the ordinal level of U.
It is most convenient to omit the quantity of the numeraire, .17.,
and its demand equation. This gives us
%
1: = Wie-+. -- -. x. se. ---. x... U). (+ # n) (48)
is ambiguously worded, since the o'er of the maximum amount of money by the con
sumer to attain given amounts of the two goods in question will change the amount
bought of all other goods. It wouldappear that the author intends that all other (than
the two goods in question and the numeraire good) goods are to be held constant, in
which case it becomes identiml with the verbal denition. The difculty may reside in
Hicks's unique use of the concept money."
Let (*Nj.) = N; be the matrm of thrs transformation Then the
element corresponding to the 1th prlce and 3th quantity represents
the literary demtron of complementanty It can casxlybe shown
that thlS, hke the rst matrix, must be symmetncal In fact. xt
15necessanly negatwe demte, hke the gbmatrn.
These last two statements are consequences of the demtlon of
the Inverse demand funct1ons accordmg to which
Nl: : kIUJln): (49)
where un,15dened to be the matru formed by omitting the kth

row and kth column from the orlgmal gbmatrlx Smce 1t 15known
from earlier chapters to be negatne demte, so too wnll be Its
nmersc
Whtle the (hagonal elements of the N and tbmatrices must agree
m sngn, being negatwe, the off diagonal terms need not agree m
Slgn except m the empirically unimportant case of three goods
Even more Important, the element m a gwen row and column of
the N mat may be of dlfferent Sign dependmg upon wh1chgood
15 called money or numeralre A sangle Instance mll sufce to
1ndxcate tlns pOSSIblllt} Suppose we have n goods m a preference
eld that can be typled by additive utlhtnes Let all but the last
good have decreasmg marglnal utxhty Then on the hterary denn
tlon the rst two goods mll be substltutes If the thxrd good xsused
as numeralre, unless the last good has decreasmg margmal utlht),
when It :s used as numeratre, the rst two goods \ull be comple
ments There rs no reason why one of the goods. whnch we may
call the last, cannot be an xnfenor good Indeed, m the mterestmg
borderlme case where one of the goods has constant margmal
utxhty, exery patr of goods lSmdepcndent on the literary demtton
when the good v.1th constant margmal Utlllty ISused as numeraire
If any other good ls used as numeralre. all other paxrs not 1molvmg
the good thh constant marginal uttllty are easrly shown to be sub
stttutes But according to the demtton of the mathematmal ap
pendlx, all peursof goods not mvolvmg the one nth constant mar
gmal utlhty must always be Independent, u hatever the numerazre
A typlcal element of N,,can be wntten out terms of the sl0pes
of the 1nd1erence locu, and the reader may verify for himself the
'pendence upon the subscript k
0
Walras was scrupulously careful in his use of the concept
numeraire. He distinguished between it and money, in the sense
of the portable, divisible, cognizable, durable, conventional com
modity which served as a counter in performing exchanges.
Marshall, when he spoke of money, ordinarily meant nothing more
than income, as we shall see in the discussion of the constancy of
the marginal utility of income. Hicks seems to alternate between
the above uses, and a third sense in which money stands for a
composite commodity made up of all but one or two of the total
number of commodities. The result is a tendency toward am
biguity, of which the above example is only one instance.
If complementarity is not of interest for its own sake, can it not
still be of importance as an indicator of conditions where certain
abnormal" phenomena are to be found? Thus, on page 71 in his
discussion of the stability of a general equilibrium exchange system,
Hicks says, If income effects can be neglected, and if no comple
mentarity is present, then the system of exchange equilibrium must
be stable." It would seem, therefore, that the concept may be
useful in indicating where a stable system will break down. Un
fortunately, this is not correct. The author has made a momentary
slip, as he has since indicated elsewhere. If income e'ects can be
neglected, then his matrix (Xn), being the sum of the negative
denite symmetrical matrices of all individuals, must itself be
negative denite and symmetrical. It must therefore be perfectly
stable on the Hicksian denition, regardless of complementarity.16
If income effects cannot be neglected, then the matrix may be
asymmetrical; according to the author asymmetry tends to make
for instability. This is not quite correct. Pure asymmetry, and
nothing elsedltends to make for stability; the neglected income
effects are the villains in the piece, and they do the most harm when
their inuence is not spent in creating asymmetry.
Again in Hicks discussion of the effects upon the "period of
production" of a change in the interest rate, the concept of com
plementarity is introduced.17 But here, too, a correct rendition of
the secondary maximum conditions shows that complementarity
"' Since these lines were rst written, I nd that Hicks has detected his error. f.
J. R. Hicks, "Consumers' Surplus and Index Numbers," Reviewof Economic Studies,
vol. IX, no. 2, p. 133, n. 2.
" Hicks, Value and Capital, p. 222, n. 1 and p. 328.
cannot be such as to make the I-Iicksian period of production de
crease With a reduced interest rate.
The Hicksian average period of production is dened as the
elasticity 3of discounted value. C. with respect to changes in the
discount rate, 3 = (1 + )". Mathematically, it is given by
(dC/dB)/C. Its precise Usefulness seems not to have been ex
plicitly indicated. Among other things, the Hicksian average
period" has the anomalous property of not being an "internal
mean," i e , one whose value lies within the limits of the greatest
and least of the time periods being averaged. Thus, it cannot be
considered a generalization of the simple Bohm-Bawerkian average
period of production. For e\ample, consider the trivially simple
"point-input-point-output" case where 99 worth of input is in
vested for one year to produce, say, $1.00 worth of output. The
Bohm-Baiierkian average period, nhiclz unlike the Hicks concept
requires a careful distinguishing between plus and mmus items, is
one year regardless of the interest rate. The Hicks period is 100
years if the interest rate is zero and innite if the interest rate is
one per cent In fact, for the so-called marginal investment it
is always innite. By introducing discount factors into the aver
age, the author hoped, perhaps, to meet the Kniglitian objection
that the period is innite, but in many cases he seems only to have
succeeded in melting the nite innite "
But this is not the place to go into the deeper Knightian objec
tions to the Austrian theory. It will be sufcient. in passing, to
state what seems to be the only essential theorem relating produc
tion planning to the interest rate. By the methods of earlier
chapters, it can be easily shown that where the rm acts to maximize
C = to + xll + x262+ + xnn) (50)
As its author indicates, here is an example of an BhsClty expressmn which is not
dimensionless For an analytic explanation of this the reader is referred to the rst
section of my chapter vx
Actually, it is not always necessary to introduce discount factors in order to make
a process With innite range yield a nite average Many ii.titers have shown how the
distant elements receive smaller weights so as to Creme a convergent innite series
Thus, imagine a perpetual stew, to which something is always being added and from
which something is always being taken out, at random Some part of what is now being
added Willnever come out of the stew, Just as some part of what is in the stew is of
innite age But it is a simple exertise in innite processes to show that the average
age of the stew is nite, and the average expectancy of a particle's staying in the stew
' nite
\!
a change in the discount factors must see the followinginequalities
realized
AxABj a o. (51)
If the interest rates for all periods are equal, and if we go to the
llmlt, we have the denite theorem
wxj/dmjafl % 0. (52)
CONSTANCY OF THE MARGINAL UTILITY OF INCOME

I return after the above digressionto an examination of another
special restrictive assumption with which the analysis of con
sumers behavior is often burdened. For its own sake the problem
of constancy of the marginal utility of income is one of the least
important in economic theory, but it has given rise to an endless
amount of discussion, most of it not on a very high level, and it
therefore deserves brief notice here. Inasmuch as I have given a
rather complete analysis of this matter elsewhere,20I do not propose
to do more than summarize. matters here, and to comment upon
some more recent contributions which became available after the
writing of the cited essay.21
As will be seen in the next section, constancy of the marginal
utility of income derives most of its importance from its relation
ships, real and alleged, to the subject of consumers surplus. In
the beginning, it is well to point out that one ambiguity runs
through the whole literature; sometimes constancy is interpreted
to mean constancy, and at other times it is intended to mean
almost constancy. The last usage generally involves some sort of
limiting process; either the change in question is supposed to be
small, whatever that might mean, or the percentage spent upon
P. A. Samuelson, Constancy of the Marginal Utility of Income," Studies in
Mathematical Economics and Econometrics: In Memory of Henry Schultz (University of
Chicago, 1942), pp. 7591.
A. Henderson, Consumer's Surplus and the Compensating Variation." Reviewof
Economic Studies, vol. VIII, no. 2 (February, 1941), pp. 117121. J. R. Hicks, The
Rehabilitation of Consumers' Surplus," same issue as above, pp. 108-116. ]. R. Hicks,
Consumers' Surplus and Index Numbers," Reviewof EconomicStudies, vol. IX, no. 2,
pp. 126-137. Robert L. Bishop, "Consumer's Surplus and Cardinal Utility," Quarterly
Journal of Economics, vol. LVII, no. 3 (May, 1943), pp. 421449.
some spec1ed good 15to be small etc , etc The problem xsstill
more complicated because the alleged result, Itself rarely worded
unambiguously, ts more often than not held to be "llkely' rather
than necessary l thmk, therefore that we shall accomplishmon.
xfm the begmmng we examme the hteral Implications of constancy
of the marginal utility of Income to show rigorously the empmcal
tmplrcatxons of thns hypothesw
We have already encountered the margmal utnhty of mcome
chapter v m the form of the ) of equatlon (31) It ISclear from
the precedmgethbnum equations (29) that the marginalutlllty
of expenditure upon each and every commodity must be the same,
equal moreover to the margrnal (rate of) utrltty denved from an
extra dollar of expendlture dtstrlbuted optlmally over all of the
goods
I have been metxculous m adhenng to the usage of marginal
utxllty of mcame, eschewmg the more common terminology en
countered the hterature of the margmal utlhty of money This
brmgs up the second fatal amblgutty mvolved m the dxscussronsof
thus subject Money has many dl'erent meanings m theoretical
dlSCLlSSlODS,
rangmg from an abstract non metallic, non paper. con
ventronal umt of account to spccrc counters, to commodttles
commandmg wzdeacceptance m exchange, to any commodity taken
arbitrarily as numerazre, to mcome or cxpendlture It ISreason
ably clear from everythmg that Marshall has written and from
the cast of l'llSthought that he demtely Intended to convey the
meaning of money sxmply as a euphemrsm for Income or expend1
turc, reckoned m pounds or dollars In hrs own words, 1t1smoney
or general purchasmg power " " In partlcular, 1t was forengnto
hrs usage to think of money m the sense of numerous, a concept
which ts not even llsted m hrs Index
This lS a matter of Importance when we come to ask the ques
teem, 'wnn respect it; 'wnatIs the nangnrau 'nu'nry( meurtre 'en
" Thxs nsanother one of the numerous places where Alfred Marshall left matters un
a fog It was part of the avor of the man not to brmg thmgs to a sharp focus BUt
what 15forgtvable In a gemus cannot be talented among lesser mortals
" Thus ls sumply one example of the \\ ong Vmer theorem discussed In chapter
n3p 34 chapter w p 66
" A Marshall Pmmples of Economic: 8th edmon p 838 For a more complete
51531011
and for detanled Citatlons the reader 13referred to my essay m the Schultz
ne
money to be held constant? _ That Pareto might have interpreted
this to mean constant marginal utility of a particular commodity
chosen as nmncraire, taken with respect to changes in the amount
of that commodity, is understandable in view of the Lausanne
usage. But that Hicks and others in the Anglo-Saxon tradition
should have thought this to be the Marshallian meaning is much
more surprising. Now at this date it is not a very signicant ques
tion as to what Marshall himself really meant, but it is of impor
tance to show the implications of at least two distinct and incon
sistent meanings. Furthermore, many writers have denitely
fallen into misstatements concerning the necessity and suiciency
of parallel indifference curves (the geometrical embodiment of the
rst of the above formulations) if various identities are to be
realized.25
A closer examination of equations (29) and (31) will show that
the marginal utility of income cannot be constant with respect to
everything. It is a function of all prices and income. If we were
to doubleall prices and income, the marginal utility of income would
have to become halved. For such a doubling of income and prices
would by the homogeneity property described in chapter v leave
all physical quantities unchanged, and hence all real marginal
utilities unchanged. But the marginal utility of income is given
by dividing each real marginal utility by the respective price.
With the numerator unchanged, and with the denominator doubled,
the whole expression must be halved. If the marginal utility of
income were constant with respect to all prices and also with
respect to income, it would have to be unaffected by a simultaneous
change in them all; since it must be halved by such a change, we
have a contradiction. Consequently, the marginal utility of in
come cannot be invariant under changes in income and each and
every price.26
At most the marginal utility of income might be independent
of all but one of these (n + 1) magnitudes. We can set n rst
partial derivatives equal to zero, but not (12+ 1). Which n shall
In addition there may be a misapprehension running through Hicks's treatment
in his book and the two articles cited. As Bishop has pointed out, the problem is not
primarily one of whether or not income e'ects n be neglected. Further, there is the
question of keeping real income constant or "adjusting" for changes in real income as
seems to be implied in certain of the Hicksian statements.
" See Studies in Illalhematfcal Economic: and Econometrics, p. 76.
we choose? Tlus can clearly be done m (n + I) alternatwe ways
One of these Involves the constancy of the margmal utnllty of la
corne Wlth respect to 71prices, but not With respect to 1ncome
Elsewhere I have argued that the xs the pure Marshalhan case
The other alternatwes mvolve the constancy of margmal utnhty
of Income nuth respect to Income and thh respect to (n - l)
prlces or n 1th reSpect to all prlces but one 7 Tlus 15the second
bypothesrs of constancy
In the Schultz volume essay I worked out the Impllcauons ol
each of these dlstmct hypotheses In terms of exerythmg that
has ever been observed concermng prlce and budgetary data the
1mphcatlons of each are hlghly unreallstlc, although 1t xsthe second
of these Wthll leads to really fantasme conclus1ons
Before summarmng these empmcal nmphcatxons I should rs1
pomt out that the marginal utility of mcome, being from a forma
pomt of we a Lagrangean mult1p11er,must by the results o
chapter v1, p 132, not be left unchanged 1fwe.go from one utlllt}
Index to another, as 15our pnvrlege m an ordlnal preference eld
At each pomt, for each d1'erent utllnty mdlcator, we have a duller
ent margmal utrllty of Income If the expressmn xsconstant for:
gwen movement when we are usmg one 1nd1cator, 1t Will not b
constant 1f we use another 1nd1cator If, to get nd of the mmu
Slgn we dene a new expressnon to be equal to ->\,xt xs reall:
necessary to mdlcate the cardmal mdex of utlhty to whnch1tapplies
We may represent the new expressmn by the letter m, and appcm
a subscrnpt to 1nd1catethe utll1ty mdex m questlon
Then,
m0 = huh. P2: I pm I) " "" X = fun, P2 : PM I) (53
As shown 1nthe Schultz volume thxs function must be homogeneou
of order mmus one, by Eulers theorem on homogeneous function
the sum of 1ts elastxcmes wrth respect to each of the mdependen
varnables must equal mmus one. so that each cannot vamsh
The good whose price changes have an effect upon the margmal unhty of mcom
need not be the commoduy selected as numeranre and It mxght prevent confusmn 1!W
made a pomt of selectmg some other commodity as money espeCIallysmce theft
no spemal reason wrthm the scope of stanc theory by tlus spemal good should b
:vrs ble durable and have all the other properties Of money Nevertheless m th
rature the Convenuon ISfollowed of pnckmg tins good as numerau'e
If we subject our utility index, go,to a monotonic transformation
to give us a new index, U = F(go), in terms of our new utility
1ndexthe marginal utility of income has the properties
1 = F,(90)m,,
dmy ,dm, ,, da
da F a + F mda- (54)
Since we can make F of any sign, the marginal utility of income

can be made to change in any direction and amount we like, except
in the special case which I have discussed elsewhere.
Have we destroyed then all possibilities of speaking of constant
marginal utility of income? Not quite, since we need only assume
that there exists some one cardinal index (even if we ourselves pre
fer to use another cardinal index, or no one cardinal index) for
which the strong, non-local restrictions of the two different hy
potheses hold.
It is a mere exercise to show that the rst, or pure Marshallian,
hypothesis implies that the income elasticity of demand for each
good must be unitary, and the price elasticity of demand for each
good in terms of its own price must equal minus one. Moreover,
the demand for each good is independent of changes in the prices
of all other goods. There must exist one way of numbering the
indifference curves so that utilities are additive and independent
in the old sense; actually except for arbitrary origin and scale the
utility function is a linear combination of terms like k.-log xi,
where the k coefcients represent the unchanging proportions spent
upon each commodity, x. These implications are both necessary
and sufcient.
In view of the well-known Engels laws and numerous budgetary
studies it is hardly necessary to point out that this ies in the face
of all observable reality, even as a rst approximation in the sense
of limitingly small changes.
Performing a scarcely more difcult exercise in connection with
the second hypothesis, it can be shown that it implies that the
demand for each good except the numeraire depends only upon its
own price in relation to the numeraire's price, and is entirely
independent of income, money or real. The demand for the
numeraire depends in a corresponding way upon all prices and
income. These conditions are both necessary and sufcient.
194 FOUNDATIONS Ul" ECUNUUC ANALYSIS
They imply among other things that the indifference varieties-are
parallel, their slopes being constant along the direction of change
in the numeraire good alone.
Of course no one has observed, and presumably no one ever
will. a preference pattern in which all of extra income is spent upon
one commodity. Note that this is not even approximately true
for instantaneous rates of change even when we neglect second
order of smalls.
Before we go into the reasons why anyone should ever have
wished the marginal utility of income to be constant, a word may
be said concerning the quasi-constancy of this magnitude. Of
course, for a very small change in the initial situation, the marginal
utility of income will change very little, as with everything elsein
the system. But this is simply a consequence of continuity, and
nothing else. It is quite another thing to say, and this is the
only important thing, that the rate of change of marginal utility of
income is small. Today the use of infinitesimals and of differ
entials is rather out of fashion with mathematicians, although a
completely rigorous basis can be given for these processes. But
in the Victorian era when Marshall was in his prime, these were
old familiar standbys, and we may be sure that the man who was
second wrangler to the great Lord Rayleigh was well at home with
their use. It is strange, therefore, that he contented himself with
gratuitous, and I believe incorrect, statements that changes in the
marginal utility of income were of the "second order of smallness."
In the technical sense they are not; the fact that we are dealing with
what can be regarded as a second derivative is not relevant here.
If it may he of comfort to anyone cherishing a fondness for
constancy, the following formula may be given indicating how the
relative amount of expenditure upon a good ties up with the clas
ticity of the marginal utility of income with respect to a change
" Perhaps King Midas was an exception, although his case involves certain dynamic
considerations outside the present scope. For a one commodity mniac of his type it
might n0t be inappmpriate to measure all Values in terms of the commodity in question:
thus, the food necessary to keep a book collector alive might be reckoned as simply me
fraction of a rst edition of Adam Smith.
" Marshall,Principfes, p. 842. The small germ of truth in the argument concerning
second order of smallness" lies in the fact that if each of the half dozen different
consumer's surplus concepts are plotted againts a variable price (later llcd 93). then
the original point, pc. they all have the common tangent. -- g..
the price of this good.
Em, _ __ k Em, Ex.- ..

EP; _ " EI ET . (5)
Providing that the two expressions in brackets do not simultane
ously become large, and thelastof these must for all goods average
out to unity, then our left-hand elasticity must go to zero in the
limit as the per cent spent upon the good in question goes to zero.
But then of course any surpluses" connected with this good
become small, and there is no reason to distinguish between the
orders of smallness.
WHY CONSUMERS SURPLUS IS SUPERFLUOUS

Of course, the problem of constancy derives its interest, I will
not say importance, from the Marshallian concept of consumers
surplus, about which an earlier generation of economists were able
to indulge in much argumentation. Since many of the points in
volved were essentially mathematical in character, and since most
of the antagonists did not go beyond literary methods, the discus
sion was not able to advance beyond a certain point, although most
of the essential difculties with the concept had been brought out.
Later when there came about a renascence of mathematical meth
ods in economic theory, economists had lost interest in the problem,
and the subject lived on primarily in the elementary textbook and
in the classroom.
In the present writer's opinion this is as it should be. The
subject is of historical and doctrinal interest, with a limited amount
of appeal as a purely mathematical puzzle. These statements are
made in cognizance of the fact that Professor Hicks has recently
attempted to rehabilitate the doctrine of consumers surplus. It
would carry us aeld from the present task to analyze his conten
tions in detail, and so only a few ex cathedm remarks must suice.
In the rst place, any judgment as to the usefulness or lack of
usefulness of consumers surplus has nothing to do with the problem
of the admissibility of welfare economics as a signicant part of
economic theory since nobody has ever argued that the latter sub
ject presupposes the validity of consumers surplus. Can it then
be said that consumer's surplus if not necessary, is nevertheless a
useful construct? Concerning this psychological question, no,u"
answer can be given. Historically the important propositions
concerning increasing and decreasing cost industries, which are
attributed to Marshalls consumers surplus notions, may be said
at best to have been incomplete derivations, and at worst may be
said to be absolutely incorrect statements which, by a pun or play
on words, seem to resemble the Pigouvian doctrine concerningin
dustries with external economies and diseconomies. In its earliei
form the Pigouvian doctrine is close to that of Marshall, but iron
the writings of Knight and Pigou himself we know that earlier four
to be quite wrong. To Pigou the problem is not at all one 0
increasing or deereasing returns, it is only a question as to whethei
each unit is taking account of its full e'ect upon social magnitude:
(other than prices), or whether it is not. If it is not, and that isal
that we mean by external economies, then there is ol course nee:
to interfere with the invisible hand. I have found nothing ii
the written work of Marshall which suggests that he ever sa
matters in this way, and even if he had stumbled upon this resull
by means of consumers surplus, it would not be the rst time tha1
a correct theorem had been reached by incorrect, heuristic reasoning
Also it may be said that the merits or demerits of the concep:
in question have little to do with the applicability of partial equi
lbrum methods to any particular problem. As for its connectior
with the theory of index numbers, after the concept has beei
renovated and altered, it is simply the economic theory of inde
numbers in the Pigou, Konus, Haberler, Staehle, Leontief, Lerner
Allen, Friscli, Wald tradition 3
F H Knight, FallaCiesin the Interpretation of Soaal Cast, QuarterlyJW"
0! Economm, XXXVIII (1924), 582-606 A C Pigou, The Economics of Walla"
4th edmon (London 1932). chap. xl and appendix iii
In the Hicks derivations (Reviewof EconomicStuds, vol IX, no 2, pp 126-137
certain well known theorems nhich are exact are derived as approtimzitions Also 11!
most interesting result, that the difference between the Laspeyre and compensatin:
variation 15equal to a generalized substitution term. is exactly true, not simply for smal
movementS. being a transcription of the familiar Othn that the two terms differ ii
consequence of the curvature of the preference eld. The one application to Ella"
coonomrcs In section 8 would be of interest only in connection With a (mlsdxrccted
attempt to measure welfare in a cardinal sense to say nhether one mot emcnt is bettei
than the sum of the benets of two other movements And even if one were interestct
ln cardinalizanon of welfare, this would not be the way to do it, for it can be shown that
the value Stuns which are used in index number theory are of importance only [Of d"
Itatlve direction 0! change which they indicate, in general (except in the triwal cast
If one were to begin afresh to give answers to the following
problems, in no one of them would consumers surplus be necessary
or desirable: Should Robinson Crusoe, a Socialist state, or a capital
ist economy build a particular bridge? Should indirect taxes be
preferred to direct taxes? Should discriminatory prices be allowed
if a uniform price will not keep an activity in business? Should the
number of rms producing differentiated products be reduced, and
in what way? Should a particular small industry be expanded or
contracted by means of tax or subsidy? etc., etc. Aside from their
extraneous inter-personal aspects, all of these questions can more
conveniently (and more honestly !) be answered in terms of the con
sumers ordinal preference eld.
It is for these reasons that my ideal Principles would not include
consumers surplus in the chapter on welfare economics except
possibly in a footnote, although in my perfect Primer the concept
might have a limited place, provided its antidote and alternatives
were included close at hand.
THE MANY FORMS OF CONSUMERS SURPLUS
The Marshallian expression of consumers surplus does not refer
to any one thing but to at least half a dozen interrelated expres
sions. It is only too easy to accept tacitly one of these as primary
and then to show that the others do not correctly measure this
magnitude. Thus, Professor Viner32 argues that Marshall is in
correct in using the area under the demand curve as an index.of
the gains from trade because this magnitude does not coincideWith
the amount which could be derived by an all or none offer, as if
0f expenditure proportionality) they nnot constitute even an arbitrary cardinal index.
Some of these implications will be seen from an applimtion of the Hicks result to what
I have called the pure Marshallian case. Let us suppose that utility really is measur
able in a cardinal sense and is given by the additive logarithmic form mentioned above,
so that demand for each good is unitary and independentof allother prices. Any two
goods will nevertheless be substitutes in the Hicksian sense-;if it seems strange that
independence in the usual objective sense nevertheless implies substitutability in the
latter sense, the reader is reminded of the fOrmal denitions laid down by Hicks. _Bc
muse each good is unaffected by a change in the other- price, a change in both Ewes
toIs'etherleads to exactly the same cardinal change in utility as the sum of'eacli c ! ":
scparately. Yet by the Hicks theorem of section 8 the combined reduction in price 'us
to a smaller gain" than the sum of the two separate "gains, an eminently gratu1
conclusion.
w York: 1937), chapter ix,
:2.- Viner, Studies in the Theory of International Trade (Ne
section iv.
193 FOUNDATIOAS OF ECONOMIC ANALYSIS
the latter were the pnmary and correct expressron for consumers
surplus As Bashop has shown," the latter may be a horse ap
prommatron than the former to the cardmal utrhty gum wluch
Blshop takes as the pnmary measure of consumer s surplus
chhs rst took the all or none magmtudc as berng the primaryone
but now adopts a fourth magmtude as pnmary, what he callsthe
compensatmg vanatton Winch rs correct can only be a quesuon
of history of thought, Much would Max-sin hate selected xicon
fronted a 1th a one or none offer? Unfortunately thrs rs no longer
an observably meanmglul question Although I think Marshall
would have agreed 1th Bishop, looking upon the other measures
as good or bad approxnuatrons, I prefer to treat no one as pnmary
or pnvnleged. but to glve the relationships between each
Thls was done m bnef but fairly exhaustwc fashion m the
Schultz \ olume essay, from whrch I append the following {en pages
copled verbatlm except for mmor changes "
Before cumming the Marshalllan concept of consumers sur
plus let us consrder the uses to wluch 1t rs put Among other
thmgs xt 15proposed as a measure of the gam (loss) of utrhty that
results from a decrease (increase) pnce of a smgle 800d A
attempt also has been made to apply ntto the analySISof the burden
Involved Commodity taxatlon It has been used to detenmm
the maximum amount of revenue that a perfectly dnscnmxmtm
monopolnst mtght exact from the consumer for a gnen amount 0
the good m questnon
Smce only an ordlml preference eld ISassumed 1n the the-or}
of consumer s behauor, there Is really httle 1mportance to be at
tached to any numerrcal measure of the gaxns from a pnce Change
In particular, one cannot fruztlully compare the gatn demed fron
a movement between tuo gwen pncc Situations wzth the gem be
tween two other prtce srtuatrons 3 Moreover. all \ahd theorem:
relating to the burden of taxatlon can be stated xndependentl)
any numerical measure of utthty change We should not be greatlf
perturbed therefore. 11'the concept of consumers surplus shoull
" Op at p 421 possum:
Pages 87-90
One can ha ever compare the garns denved from a change m the basrc Pr
sutuauon wlth an alternatwe pnce change from the some baszc sntuauon smce
resolves Itself into an ord nal companson of the alternatne new sltuatrons The mate
Jatlon cancels out so to speak
be found to be inadmissible. Its only advantage seems to lie in
its deceptively easy two-dimensional representation.
Consider an initial price and income situation, (pf, . - -, pa, I),
and the corresponding amount of goods purchased, (x10,- - -, x,,o),
For any selected utility index, go,there will also be a given amount
of utility, ,a(X). Suppose that a change is made in but one price,
{>,-,and income is left unchanged. There will be new amounts of
every commodity, (xlb, - - -, x,,b), and of utility, <p(X"),correspond
ing to the new prices and income, (1915,- - -, 193,1), or (pf, - -
12,-",- - -, pu, I ), where {).-bis less than pi.
We are interested in the following magnitudes:
1. The gain (loss) in utility resulting from the price change, or
@(X")- MX)
2. The area between the demand curve of the ith good and the
p.-axis Within the range of the price movement, or
P.
_ f p a xidpio
3. The amount by which the expenditure on the 13thgood in
the new situation is exceeded by the maximum amount of money
which the consumer would be willing to pay for x.-"in preference
to trading at the old set of prices. (This may be negative if we
are dealing with a price increase rather than a decrease.) Call
this bEab.
4. The amount of extra income which the consumer would in
sist upon if he is to be as well o' as in the new situation while
consuming the old amount of xi. Call this Eab.
5. The change in income which will make trading at the new
set of prices as attractive as trading at the old set of prices with
the initial income. Call this bAIab.
. The change in income which will make trading at the old set
of prices as attractive as trading at the new set of prices with the
initial income. Call this Aha.
According to the Marshallian doctrine of consumer's surplus,
all six of these magnitudes are equal except for dimensional con
Interchanging subscripts changes algebraic signs. Thus,
Alas = MI. AI, but not "AIabnor 5E, can exceed I.
stants We are expllcxtly warned, however. that hxsdoctrme holds
unqualxedly only when the marginal utnllty of mcome ISconstant
and only If utllmes are independent I shall now examme the
value of each of these magnxtudes four cases (a) m the general
unrestncted case of Stable demand, (b) under the rst mterpreta
non of constancy of the margmal utxlxty of Income, (c) under the
second hypothcsns when the 1th good 15 not the numerazre, and
(d) under the second hypothesns when the zth good Itself has con
stant margnnal utlhty of meome Only the most sketchy proofs
w1llbe Indicated
In the general case we have the followxngrelations
- -f,()f()
=jptm.$(p,-f)dpl= _ffmgdp. (56)
AI... = max (2"; pf'xf Z:p,"x,), where qa(X) = rp(X). (57)
1 z
IW P:b(x:b _ x)): (58)
EW - Mac; (591
If only the zth pnce changes thxs becomes

bAIab (Pt "' 939333. (60]
Slmllarly,
a ab
mm 0:: m, -
)::prxr) where par) = saw) (611
.AIab M ww ae,-). (621

l
and, the 1th puce alone changmg,
AIab P:(x: " x1) = (P.. pi xob (63]

I
Although this is not the Marshallian interpretation, consumer'ssur
plus seems to be most justied in this case. However, the above
equalities cannot hold simultaneously for every good.
For ease (d) equation (67) must hold with the possibleexception
of the inequality referred to in a previous footnote, but (69) is
denitely false.
Other results may be extended in at least two directions. Mort
than one price may be permitted to change. and also income,with
out changing many of the inequalities This may be left to the
interested reader as an exercise. Also we may attempt to deriw
rigorous inequalities in the n commodity case. This offerscon
sidernble complexities in connection with concepts (3) and (4).
Case (c) rs :uimnt to insure the equalities oi equauon (69) Some oi them ma:
hold m other cases
ln general. lme mtegrals Willreplace sunple Integrals Nth the path of integratxor
oi the former a matter of no consequence
CHAPTER VIII
WELFARE ECONOMICS
BEGINNINGas it did in the writings of philosophers, theologians,

pamphleteers, special pleaders, and reformers, economics has al
ways been concerned with problems of public policy and welfare.
And at least from the time of the physiocrats and Adam Smith
there has never been absent from the main body of economiclitera
ture the feeling that in some sense perfect competition represented
an optimal situation. Of course, over time the exact form of this
doctrine has undergone modication (not always in any one direc
tion), and there is considerable diversity in the attempted proofs
(in the amazingly few places where rigorous proof was attempted).
Although this doctrine is often thought to be conservative or
reactionary in its implication and to reect the kept" status of
the economist, it is important to emphasize that it was radical"
in the eighteenth century, and there is some evidence from events
of the last decades (e.g., the T.N.E.C. and economists role and
views with respect to Anti-Trust) that it has become a thorn in
the side of what are usually thought of as conservative interests.
Furthermore, some Socialist writers, who in their youth became
interested in analytical economics, nd in this doctrine a possible
device for expediting planning in a socialized state.
Early uncritical allegiance to this doctrine arose in part from
the understandable eighteenth-century tendency to nd teleological
signicance in the workings of what is after all an equilibrium sys
tem which is not devoid of aesthetic content regarded simply as a
mechanism.1 But it would be unfair to the older economists to
believe that their case ended with a simple argument from design,
1It would be out of place here to discuss the relationship of this doctrine to that of
natural rights"; to that of competition as an immutable law with which man cannot
interfere even if he should wish to; to the inverted doctrine of natural selection whereby
the results of competition were judged to be best by means of a circular denition of the
ttest" as those who survive; to the Malthusian view that hardship and competition
are necessary to bring out the best" in a man; to the view that competition was good
enough for our predecessors and therefore good enough for US:and ' "5
designed to preserve the status quo.
203
204 FOUADATIONS OF ECONOMIC ANALISIS
even If such a charge can be sustamed \\ 1th respect to certam
Ep1g0n1
This can best be seen an the untmgs on Internatnonal Trade
where the Issue of the tanffbnngs out most clearly the u elfare and
policy behefs of economists, eyen down to the present day For
free trade rs but one (lramatlc exemphcatlon of pure competition,
and m tlus eld formal attempts at proof were made, or uc can at
least In many cases plece together the tmphcrt beliefs of the author
1 Perhaps the most common reason for behevmg competntnon
to be optimal stemmed from the recogmtnonthat no party could be
hurt by exchange as compared to has posmon before trade. smce
he could ale.ays refuse to trade Thus, trade 15better than no trade,
exchangets mutually beneczal, one party does not gam aka! the other
loses If e examine the argument carefully, we nd that It does
not really tmply that pure competttton IS optrmal, even though
properly Interpreted It can provrde a case against prohzbmvctans
2 A second more sophlstlcated argument, Wthh mcludcs the
rst and more rests upon the fact that the equlhbnum pos1tlon
reached m pure competxtlon represents an optmmm for each 1nd1
\ xdual conSIStent\\ 1th hlSongmal endowment of commodities and
the market s1tuatlon wrth which he 18confronted But ever) mdl
udual may be malung the best of a Sttuatton mthout nnplymg
that that best lS \ery good or ISoptimal, although each 1ndw1dual
pure competmon takes pnce as gnen, for the market It ISa
varlable, and 1t ns qmte possnble that condxtrons other than pure
competltron m1ght lead to better results In terms of any of the
usual ethical notions But, leavmg asade all ethlcal notions, as it
not equally clear that under (say) monopol}, both bu) er and seller
are domg the best that they can for themselxes under extstmg
market Situatlons> The only dlstmgulshmg feature of pure com
petmon, as compared to any other mode of behavnor, IS that the
market condltlons facmg each mdtudual are taken (by hlm) to be
stralght lmes" rm olvmg trade at unchangmg pnce ratios And
It IS precnsely the questnon of the sense 1n whlch thlS IS optimal
whrch ISleft unansn cred
It does not appear that Walras exer reached beyond thls second
ge of the argument Hts cardmal fazhng CODSlStS
not so much
compare the very penetrating remarks of \Vlcksell on this pomt l\ WcLscll
I 724183on
83 Pohtrcat Economy (Engl sh translation New York Maumllan, 1934)
WELFARE ECONOMICS 205
in the fact that'he jumps from incomplete premises to sweeping
conclusmns, but m the fact that he is satised with this very limited
kind of an optimum, which by a play on words he seems to confuse
With the more usual and important senses in which perfect compe
tition is conceived to be optimal.3
3. Still a third stage of reasoning attempts to show, not that
each 1nd1v1dual IS made best o' by competition since this is im
p0551bleunless each can take all, but that in some sense the sum
total of satisfactions is maximized, that perfect competition effects
an ideal compromise of mutual benet, or in its most nebulous
form that free trade (perfect competition) maximizes world (all
individuals) income. Of course, this involves the notion of adding
the utilities of different individuals, of somehow being able to com
pare and weight the utilities of different individuals. Although
the marginal utility economists, with the exception of Jevons who
made an interesting slip in connection with the concept of the
utility of the trading body, knew that it was not necessary to
make interpersonal comparisons of utility in describing exchange
under pure competition, they nevertheless did not have the modern
reticence about making such assumptions.
Launhardt seems to have been the only economist who at
tempted to give rigorous proof of this theorem. As VVicksellhas
pointed out in the section just cited, his argument is mathematically
and logically false. Yet he must be given credit for having made
an attempt at rigor, and we can learn more from his unambiguous
failure than from many pages of fuzzy literary effusion.
To many modern economists the diiculty with this third line
of reasoning lies in the fact that it assumes that the utilities of
different individuals can be compared, in fact added together.
If interpreted literally, he would seem to imply that each and every person is made
better off by perfect competition, a conclusion which, as \Vicksell observes, goes farther
than the free traders themselves, for the latter have not denied thata restriction of free
competition might be most advantageous to a small privileged minority." Ibid., P- 76
Actually there is one qualication in Walras' argument which makes it not so much
wrong as trivial. He maintains that perfect competition creates a maximum .of satis
faction, consistent with trading at uniform prices. \Vaiving the trivial objection that
under nondiscriminating monopoly trading is also done at uniform prices, I find this
confusing. Except for positions of multiple cquilibria which we may provxsmnally
ignore, the equilibrium position under competition is uniquely determined. Instead of
being the optimum condition under these conditions, it is the only one possible. Thus,
it is the worst position as well as the best.
206 IOUNDA TIONS OF ECONOMIC ANALYSIS
This they would regard as unscientic." But to the preceding
generation of economists, interintlividual comparisons of utility
were made almost iiithout question; to a man like Edgeworth,
steeped as he was in the Utilitarian tradition, individual utility-
nay social utilitywas as real as his morning jam. And with
Marshall the apostrophe in consumers surplus was always after
the s.
Both Marshall and Wicksell ObJCthd to what they considered
to be a prevalent notion that perfect competition leads to the
maximum of satisfaction Both enter as a minor objection the fact
that there may be multiple positions of equilibrium; actually this
is largely irrelevant since each stable equilibrium might be a rela
tive maximum as compared to points in its immediate neighbor
hood (rm klemen) even if it were not the maximum maximorum
But their mayor objection consists in the fact that with existing
distributions of wealth and ability, the processes of imputation
under competition Willgive rise to great inequalities in the personal
distribution of income so that unless individuals are very different
in their natures the marginal utilities of income will not be equal
for each individual Both recognize that in these circumstances
any interference ( la Robin Hood) with perfect competition which
transfers income from rich to poor would be benecial.
4. It might be thought that at this stage Marshall and \Vicksell
would enunciate a fourth proposition, that exchange under perfect
competition is optimal provided the distribution of income is
optimal. In the case of Wicksell the proof which he gives (Lectures.
p. 80) to show why perfect competition is not optimal when the
distribution of income is inappropriate paves the way for a proof
as to why perfect competition is optimal when the distribution of
income is appropriates1l \Vicksell also realizes that when the dis
tribution of income is not optimal, the creation of a condition of
Actually his p'roofseems to suffer from one minor dm back In e'ect. his evalua
tion of the change in utility resulting from a change in price Irom the competitive level
assumes that in the noncompetitive Situation all indiVidualsare still on their o'ercurves
Strictly Speaking, this is not possxble It would perhaps be c0rrect to say that his 1111'001
(With slight modications) shows that transfer of goods or mcome from one mdmdual
to another could not improve the competitive conditions There is also an unfortunate
minor slip in expresswn, perhaps in translation. in the statement that free competition
would secure a maximum satisfaction to all parties to the exchange " (Ibid . Po31. my
' C5) Actually, It is the sum of all and not the utility of each that is maximized.
imperfect competition may improve the situation, but that this
lS not the best way of improving the situation, since perfect com
petition IS a necessary condition to maximize production." I
return to this point a little later in this historical review.
Although inappropriateness in distribution is thought by Mar
shall to render the competitive position suspect, he was of the
belief that many decisions involve alternatives which affect all
classes more or less equally. He has been criticized for this too
facile assumption, but it is nevertheless true that many modern
economists, and this includes some purists, by use of the principle
of sufcient reason (or is it insufcient reason ?) argue in such terms
for or against a price change of a commodity which is not presumed
to relate more to the poor than to the rich.
However, aside from problems raised by the inappropriateness
of the distribution of income, Marshall had important objections
to the equilibrium position realized under perfect competition.
These objections resulted from his analysis of consumers surplus,
an analysis which was regarded as almost the most signicant
contribution of his Principles. By a comparison of geometrical
areas he arrived at the conclusion that increasing cost industries
would be pushed to too great a margin under competition, and that
the output of decreasing cost industries would be too small under
competition. From the modern standpoint it is clear that these
conclusions are true in only a very limited sense. And if Marshall
did arrive at conclusions which are not completely wrong, it is
nevertheless clear that he arrived at them for the wrong reasons.
It is not easy in a few paragraphs to delineate the various faults
in the Marshallian reasoning. In the rst place, his exposition in
Book V, chapter xiii, is extremely sketchy, and, in the second
place, it is impossible to avoid the somewhat extraneous difculties
arising from the admittedly unsatisfactory treatment of decreasing
cost by Marshall. However, the latter is simply the most dramatic
exemplication of the paradox that Marshall, with whom the doc
trines of partial equilibrium and industry analysis are inseparably
associated, nowhere presents a complete or satisfactory theory of
the industry in its relationship to the rms which make it up. If
anyone doubts this, he need only compare the treatment of these
problems by Pigou, Marshalls shining pupil, in his 1912 Wealth
and Welfare with his treatment in the late editions of the Economics
208 FOUNDATIONS OF ECONOVIC ANALISIS
of Welfare or thh that of Vmer tn the Zeztsclm article (71th 1n
chapter ru
Another madequacy but one that can be easrly remedled hes
m the fact that Marshall neglects producers surplus Instead of
treating this symmetrically ulth consumers surplus so that rt rs
possrble by the reasomng of page 468 footnote 2 to amve at the
curious conclusron that mdnstrzes of mcreasnng costs should In
many cases be contracted men If there are no dccrcasmg cost
moustrles to expand
There seems to be no pomt duscussmgthe \hrshalhan reason
mg at greater length except to note that IIgou nowhere makes
essentnal use of the concept of consumers surplus 1n lus \\ell'are
amlysus He onglmlly enuncxlth essentially the Marshalhan
conditions \\ 1th respect to uncreasxng and decreasmg cost mdus
trues but as a result of the crittersmsof All) n Young nght and
Robertson he scnously modzed these conclusnons In 1118nal
form lus doetrme holds tlnt the equzhbnum of '1 closed economy
under competatron nscorrect cwept here there are technological
external economies or (hsecononues Under these conditions sance
each mdxvrdual s actions haVe effects on others Inch he does not
take Into account ln making lus decnsron there ns a pnma jam
case for Interentlon But thls holds only for technological factors
(smoke nursance etc ) changes {actor pnces resulting from the
expansnon oi demand by rms m an mdustr} represent transfers
whlch are Irrelevant for determination of Ideal output (In all
lazrness It should be admntted that the correct use ol consumers
surplus and producers surplus might h'm. helped to av01d error
m thts regard )
There \\ ould be no reason to rake oxer these old ashes uere 1t
not for the fact that Professor Hicks Ins recently lent the u eight
of his authonty to the \ IC that the doctrme of consumers surplus
has a dorm to rmportance the \\ elhre eld As I have mducated
chapter \ n caret'ul'perusal of l'ns argument Simply conrms my
' F Il kn ght Fallac es m the Interpretation of Soc al Cost QuarterlyJournal
of Econom(: (I923) Repr nted n TheDh sof Camps!: on (\conrk "et? 1935)
pp 215 236 Allyn Young I e__ouslth and . cllare QW'C'I Journal of
Fconamtcs xxvn (1913) 672-686 D !! Robert on Those Empty Boxes Eco
nam cJournaI XYXH (1924) 1631
. R H cks The Rehabltat on of Consumers Surplus Rez-moOf Econo mc
sruam mn (1941) nosno
belief that the economistmathematical, literary, beginner, expert
hadbest dispense with consumers surplus. It is a tool which
can be used only by one who can get along without its use, and
not by all such. As Hicks admits, it is not useful in the exposition
of the conditions of equilibrium or optimum. And even in
the case of a Crusoe economy where the problems raised by many
individuals can be sidestepped, it is usually only devised to give
the loss in utility resulting from a deviation from the optimum in
the amount of one good.
In this connection its principal conclusion states that the
(second order) change in utility resulting from a deviation in the
amount of one commodity, other commodities continuing to be
optimally adjusted, depends upon the amount of the discrepancy
in that good times the discrepancy in the equilibrium condition.
This conclusion is no more plausibly deriyed from consumers sur
plus than from simple intuition. And if one probes deeper, one
nds in any case that the theorem is incorrect even to the order of
innitesimals (second) at which the argument is pitched.
Thus in the most favorable case to consumers surplus where one
commodity, xhas literally constant marginal utility so that
U = L(x1, xe, , xn) + mxn+1, (1)
and where the goods can be converted into each other at constant
technological rates as indicated by the relation
n+1
z (J,-x,-: C, (2)
1
for this conclusion to be correct it would be necessary that the

change in utility resulting from a small change in the amount
of xl be
5217 = 0 + %L116x12. (3)
Actually, by a simple extension of the reasoning in chapter iii, it is

given by
52U = 0 + % : Lijxixj- (4)
1 1
If we proceed to higher orders of innitesimals, the case is worse,

and the same can be said if we drop the unrealistic assumption
210 FOUNDATION? OF ECONOMIC ANALlSIS
about the utrhty of the numerafrc and If the ongmal posrtron 15
not one of equnhbrrum
Even If consumers surplus did gne a cardinal measure of the
change m utility from a gn en change xt 15hard to see what use
thus could serve Only m the contemplatron of alternatwe move
ments whxch begin or end m the same pount could tlns cardinal
measure haw any sngmcance and then only because xt ISan mdl
cator of ordinal preference Such s1tuatrons are comparatnely
rare as far as questions of socnl pohcy are concerned bcmg m the
nature of the somewhat academic question as to whether the Intro
ductlon of a httle monopoly exll mto one industry all others bemg
competrtnc before and after the change xs better or worse than
mtroduetron of some monopoly mother
In connection 1th monOpohstxccompetrtlon the frequent oc
currenee of decreasmg cost and mdwrsrble mmal costs mentably
rarses problems of an all or none character Wan mg the drf'
cultres ansmg from many and rduals ne see that correct decrswns
necessrtate reference to ordrnal Indifference cun es and to nothing
else Certain difcultles connected u nth the determrnatron of the
optimal amount of dlfferentratxonof product u ere properly posedby
Chamberhn Cassels and Kahn \uthout the use of consumers
surplus
We may conclude from tlus lengthy digressnonthat after makmg
due allowances for external economies and for certain onussnonsm
thenr exposrtlons the founders of neo classrcal economrcs behexed
that perfect competltlon led to an optimum m exchange and
production prov1ded that the distributron of 1ncomeu as appro
pnate But they dld not behexe that Incomes Imputed by the
competmve process as of a gn en hrstoncal dlstnbutron of owner
shlp of factors of productxon and personal abnhtzeswas nnany sense
the best one and not subject to modrcatlon by appropriate
mechanisms
5 Before analyzmg the problems encountered under the head
mgs of optnmum production and optxmum exchange I should
hke to note brrey the exlstence of econormsts \\ ho attempted t
' E H Chamberl n The Theory of Monopa! mc Compem on (3rd ed Carnb dge
Harvard Un vers ty Press 1938) p 94 j M Cassel: Excess Capaerty and Monopo'
13t1c Compent on Quarterly Journal of Ecanarmcs LI (1937) 426443 R F Bah"
Some Notes on Ideal Output Econam c Jaume! XLV (1935) l 35
WELFARE ECONOMICS 21 I
establish the stronger position that incomes imputed under compe

tition were actually right and best. Thus at an earlier date,
Bastiat, whose powers of analysis were hardly of the highest order
even in his day, hoped to show that benecient competition would
lead to . . . an amount of utility and enjoyment, always
greater, and more and more equally distributed. . . ."
Confronted with the undeniable fact of considerable inequality
of income and possessing the latent Western European prejudice
against inequality, writers had either to refer to a future day when
competition would achieve better results, or attribute existing in
equalities to the admittedly large institutional deviations from
competition, or to look for inequalities among individuals' charac
teristics (including property ownership) to justify differences in
income.
To anyone with knowledge of the world the perverse relation
ship between exertion and income made necessary a revision of the
classical real cost doctrine in its simplest form, although the promo
tion by Senior of abstinence to the rank of a full-bodied real cost
helped to bolster that doctrine. But ultimately refuge was found
in the undeniable fact of differences in personal ability" and the
related doctrine of non-competing groups. This raised many ques
tions as to the extent to which the relevant abilities were or were
not acquired characteristics and the degree of correspondence
between the distribution of abilities and income. That much of
this discussion was meaningless and from most points of view ir
relevant does not detract from its signicancefrom the standpoint
of the history of ideas.
Among analytical economists J. B. Clark9 is best known for
his belief that not only will factors of production have imputed to
them their marginal productivity under competition, but that this
is a natural law which is morally justiable" since this is their
actual," specic" product. Indeed Clark himself considered
that the principal way that his independently discovered marginal
productivity doctrine represented an improvement upon von
Thnen lay in his demonstration of its ethical fairness as compared
to the latter's belief that the doctrine involvedexploitation. That
Clark, who clearly states the distinction between personal and
' F. Bastiat, Harmonies of Political Economy (2nd ed.; Edinburgh, 1880), p.
]. B. Clark, The Distribution of Wealth (New York, 1899).
212 FOUNDATIONS OF ECONOMIC ANALlrSIS
functronal Income should have thought that he had proved the
ethrcal farmcss of Income determrnatron under competrtron rs
srmply a reection of the fact that where emotronal behefs rn right
and wrong enter rnto analysrs rt rs usually not to the advantage of
the latter As e shall see even rf all rneome resulted from per
sonal servrces Clark s proposrtron rs not consrstent wrth wrdely
held ethrcal \ news and 1frt rs accepted as a dcmtron rather than
a theorem rt wrll be found to be consrstent wrth no unambiguous
ethical evaluatron of dr'erent Indrvrduals welfare Nevertheless
rt has consrderable appeal especrally m a {rentrer scelety where
each Indrvrdual could be thought of as uorkrng by hrmself under
condrtrons where hrs product could be rdentred Analytrcally
rt rs almost precrsely these terms that Clark rst percerved hrs
doctrrne gorng wrth parnful slowness from the (broad) zone of
rndrfference to the concept of the rnternal margrn
6 WhaleWleLselland Marshall held that eompetxtron would be
optrmal If the dlstnbutlon of mcome ere approprrate It was left
for Pareto ' to take the stronger posrtron that competrtron pro
duces a maxzmum d amaze collecme regardless of the drstrrbutron ol
mcome and mdeed even 1fthe utrlrtres of drfferent rndrvrduals were
not consrdered to be comparable An optrmum pasthou m tins
sense was dened by the regmrement that there should not em! any
passable umatum or morement winch would make everybodybetter of
Hrs drscussron rs not easy to follow and 1t has not recerr.ed
attentron from economrsts commensurate wrth the Importance
whrch he attached to rt Yet rt forms the basrs of many modern
notrons and rt led drreetly to the Important contrrbutron of Barone
Pareto also seems to have been one of the rst to drscuss cnterra
of plannrng under collectrvrsm
Pareto s exposmon IS comphcated by the fact that he works
wrth drfferentrals or rst order (rnnrtesrmal) varratrons This was
a very common practrce wrth mathematrcrans and physrcrsts of
the nrneteenth century and because of rts formal heunstrc con
venrenees rs strll often used today And under proper qualica
trons thrs praetrce can be gwen a ngorous unambrguous basns
Nevertheless u here delrcate problems of rnterpretatron are m
volved rt often obscures more than 1treveals eSpecrally 1fthe prob
"V Pareto ManuelD Emma Polttrque(1909) chap v also the Mathematml
Append x par 89 pamm
lem arises as to whether any given differential expression is an
exact" differential.
Pareto is unwilling to add the utility or ophelz'mz'te'of different
individuals together, either in toto, i.e. ( U"l+ U2 + - - -), or for
small variations (5 U1 + U2 + - -). _For that would involve a
comparison of different individuals utility, and in addition it would
depend upon the particular cardinal index of oplzelimz'te selected for
each. But he was interested in comparing the summed variation
in the utility of each, after these expressions have been divided by
the marginal utility of any one good, a, selected as numeraire.
For if we examine the dimensionality of the expression
W1 5U 1
+ U2 + ---, (S)
where as usual subscripts represent partial differentiation, but
where the superscript is taken to indicate different individuals, we
shall see that this has the dimensionality of the good, 0, and
nothing else.
Pareto attempts to show that if the original position is one of
equilibrium under perfect competition, then no possible variations,
consistent with the fundamental scarcity of goods and given tech
nology, can make the above expression positive. If it could, he
says, it would be possible to arrange things so that each term in
the expression could be made positive, and then everyone would
be better off. But the expression cannot be made positive. Ac
tually, regarded as a differential expression (of the rst order), the
above expression can be shown to be zero, in consequence of the
fact that each ,commodity is sold at minimum unit cost (propor
tionality of marginal products, etc.,11and in consequence of the
tangency of each individuals indifference curves to mutual price
exchange loci. If recourse is had to differentials of higher order,
the secondary maximum-minimum conditions of rms and indi
viduals will guarantee that the expression (5) will be negative for
all nite deviations from the competitive position, or so Pareto
attempted to show. _
Although Paretos treatment is somewhat sketchy and in need
of expansion, part of which Barone later providedn the mam out
lines are reasonably clear. But in connection With the funda
u Ibid., p. 646.
mentals of Interpretation of the Slgmcance of his maxxmum, there
do anse certam problems Fu'st, can the dlfferentlal e\pre5510nof
(S) be regarded as the exact drerentlal of some etpressmn? Ac
tually, Pareto later glves a name to th1s expressmn, calling 1t U,
but IS there an expressmn U (soc1al utlhty?) of whtch thrs 15an
exact differentlal? Pareto does not tell us, but presumably he
would answer. no, if he were on guard when the questlon was asked
As we shall see, Barone does work 1th an expresswn whose dl'er
entxal corresponds to (5), but he clearly recogmzes that 1t xs a
constmctron, not Im.olvmg the dtmenswn of utlhty, but rather
that of the numeralre good
But the most Important objection to Paretos exposntlon 13his
lack of emphasxsupon the fact that an optlmum pomt, lus sense
:s not a umque pomt If transfers of Income from one 1nd1v1dual
to another are arb1trar11y 1mposed, there Will be a new optimum
pomt, and there 13absolutely no way of decxdmg whether the new
pomt lSbetter or worse than the old H15optimum pomts const:
tute a mamfold mmty of values Thus locus can be obtamed
under reglmes qmte dlfferent from perfect competxtxon (e g, by
multilateral monopoly) \Vnthln Pareto's system 1t 151mpossnble
to decxde, by hlS dl'erentlal cntenon or otherwnse. which of two
pomts on what may be called the generaltzed contract locus" are
better, or even that a gwen movement off the contract locus and
hence to a non optimal poxnt 15good or bad Actually m terms
of the under reference schemes of ordmary economlc thought such
a movement may be deemed emmently desnrable But Pareto
shows that however deszrable such a move may be, there emsts st1ll
a better move, which for the same (ordmal) amount of harm to
those Who should" be harmed. Wlll yzeld more benet for the
worthy ones who are to be beneted Thls xs an 1mportant
contrlbutlon
7 1.9a masterly artmke, mrrttas- m 1503 .? kalam bat m!
translated mto Enghsh unt 1935,laBarone developed further and
m greater detanl the Paretlan condltrons of optimum. especnallyas
In hls rher Cour: dtscussxon II 90 H he exphcntly assumes the dustnbutmn of
Income to be convenable but m 1118later Manuel treatment the dependence of the
optimum pomt on the mmal dustnbutxon of Income and hence Ita lack of umqueness
:s not brought to the foreground
" Reprmted as an Appendlx F A Hayek ed Calledwut Economic Ptannml
p rlge 1935) pp 245-290
they relate to the planning of production under collectivism. By
avoiding all mention of utility and indeed without introducing even
the notion of indifference curves, Barone was able to break new
ground along lines which have in recent years become associated
with the economic theory of index numbers.
Unlike most of the writers discussed above, Barone is unsatised
with the statement that free competition maximizes product, or
sums of product, which can then be distributed in any given
fashion. Heterogeneous products cannot be added. Further
more, leisure may be preferred to the maximization of output. It
is signicant that those writers who do not explicitly introduce the
equations of general equilibrium should gloss over the denition of
product which is allegedly maximized. Thus, VVicksell con
nes his demonstration to a case where the same product can be
provided by different sources, and only in this case shows that
various marginal conditions are optimal. The same is true of the
very excellent treatment by Knight, in which the movement of
goods over alternative roads is analyzed for optimum conditions.15
It is remarkable that Professor Pigou, who reaches substantially
correct conclusions, never squarely meets the problem of the deni
tion of social product. His index number discussion represents an
important contribution in its own right, but it is offered at best as
an approximate criterion or indicator of changes in individual and
social welfare. He would not seriously suggest that the thing to
be maximized is the money value of output deated by an ideal
index of prices. Nor will the more exact limits of index number
theory such as are discussed in chapter vi avail.
Barone proposes to add different products after they have been
weighted by their respective prices; it is usually taken to be con
venient to express these prices as ratios to the numeraire good, a.
For Barone, productive services can be treated simply as alge
braically negative goods and services. Thus, decisions between
more or less work can be included in his welfare system. Then if
the sum total of each good consumed by all individuals together is
written as
A=+w+~o
B=w+m+~- @
etc.,
Wicksell, op. cit., p. 140, passim.
Knight, op. cit., p. 219.
216 FOUNDATIONS OF ECONOIIHC ANALYSIS
1nd recalling that
6U _ Uaa + Uob + (7)
we can write Pareto 5 equation (S) above m the equivalent form
1(60+ 6a+ ) + mob! + W + )+
=M+mw+ @
In passmg from (4) to (8) use rs made of the fact that the ratios of
marginal utllltleSof two goods for each Inleldual are equal to their
price ratios Barone himself does not use this terminology but
no doubt he would haxe to If the connection wrth Pareto were to
be shown
The expressron in (8) can be regarded as the variation in the
followmg expressmn \\ hen pnces are regarded as constants
=A+pbB+ (9)
If productive servrceswere known to be constant so that they could
be neglected (9) would equal (except for dynamic factors mvolvmg
capital which we can ignore) money value of national product If
all productive factors are included it Willrepresent the net differ
once bet een value of consumers goods and the return to produc
tive serVices Under many assumptions this quantity must be
zero under the full conditions of perfect competition
Barons shows that perfect competition maximizes this expres
sxon prrces being taken as xed parameters 1e any variation from
't condition of price equal to minimum cost must make <I> as gwen
m (8) negative Thus if we are at conditions other than perfect
competition With (8) not equal to zero for all posszble variations
it 111be possrble to spec1fy a movement which Will make 54p031
tlve But we can think of 64>as being made up of the sum of a
Similar expresswn referring to each indiVidual
54 = W + 640+ = (601 + Pbbl + )
+ (sa2 + p,,b+ ) + (10)
If for any movement the total 64>is pOSlthe it 15not necessary
that each and every one of the mdn idual 5 ap be posrtive but it
5 necessary that those which are posrtrve should outweight those
wlnch are negative Thus those who are hurt could be compcn
sated by those who are helped, and there would still be a net gain
left to be parceled out among the individuals.
This is essentially the gist of the Barone argument. The one
point which will occur to the critical reader is the fact that arbi
trary prices are assumed in evaluating the expression to be maxim
ized. Which prices are to be used? Barone employs the prices
which prevailed before a contemplated break with the conditions
of competition are made, and this sufces if one merely wishes to
demonstrate that not all individuals can be improved by any
departure from competition.3
Unlike Pareto, Barone satises himself with deriving optimal
production conditions without going into the fact that under compe
tition no additional individual exchanges of xed amounts of goods
would be mutually protable. No doubt this oversight resulted
from his wish to avoid the use of indifference curves and utility,
but even without these constructions, by the use of the index num
ber notions which he pioneered, the enlarged conditions of exchange
could have been included. It is a tribute to this work that a third
of a century after it was written there is no better statement of the
problem in the English language to which the attention of students
may be turned.
8. The next writer who deserves our attention is A. P. Lerner,
who comparatively recently developed, presumably independently,
the Paretian conditions which show that the marginal equivalences
realized by perfect competition lead to an optimum of production
and exchange in the special senses discussed above.17 Actually, in
the eld of production his statement of the problem is slightly
different from that of Pareto and Barone. They showed that a
movement to conditions of perfect competition in the eld of pro
duction and cost could make everyone better off because they could
be given more of every good. But they still worked with indi
viduals. Even in a collectivized state where the existence of the
individual is not assumed, the Lerner formulation of the sense in
which output is optimal would still hold: the marginal equivalences
of competition are such as to give a maximum of any one product for
" Actually, Barone discusses varying prices in a passage which seems obscure to me.
0p. cit., p. 255.
" A. P. Lerner, The Concept of Monopoly and the Measure of Monopoly Power,"
Review of Economic Studies, I (1934), 157-175. Economic Theory 'and Socialist
Economy," Review of Economic Studies, II (1934), 51-61.
given specified amounts of all others. This is almost identical with

the Pareto-Barone production propositions, but not quite.
Professor Hotclling. also presumably independently. developed
in two articles " conditions closely related to the Pareto production
and exchange conditions of optimum In particular. he has n
sistcd upon the fact that marginal rather than average costs provide
the appropriate liasis for pricing. and he (lexeloped the dramatic
applications of this to the problem of railroad rates and decreasing
cost public utilities in general On the analytic side each thing
\\lnch he sets out to prove he does prove it itli great elegance and
generality, but his fundamental primitive assumptions are only
impIiCitly related to each other and to the equations of general
equilibrium. Moreoxer. his neifare \\ork really falls into mo dis
tinct headings, on the one hand. that of the first article cited and
much of the second article. and on the other hand, that of the
second section of his second article referring to the "fundamental
theorem" (especially pp 248256). Roughly these two (lit erse
contributions of iiotelling fail respectively under the headings ol
optimal production and optimal exchange conditions; or, on the
analytic side, to the difference between firms \tith unlimited
budgets and the consumer with limited budget. to each of nhich
elds Professor IIotelling has contributed much in the itay ol
demand analysis. This dualism explains uhy so discerning a
reader as Professor Frisch should hate been puzzled by the
Hotclling proof "
" H Hotelling, The Genenl \\ elfare in Relition to Problems of Taxation and of
Radium)and Utility Rates. Ixonometnra. \ l (i938). 242-269. Edge orth's Taxation
Paradox and the Nature of Demand and Supply functions." Journal of Politica! Econ
omy. XL (1932), 577-616
" Space cannot permit a detailed examination of the exact steps in the Hotelhng
reasoning, this being particularly unnecessary since it is clear that his eoncluswns an!
impeccable In the original speczuon of his system Professor liotclhng essentially
generalizes the DupuIt-Marshall partial equdibriurn set up to many interrelated indus
tries However. unless we conne oursels es to the production problem alone. this I
not lend to the equatwns of general equilibrium "lhose require the addition ol the
speaal demand functions of consumersfor goodsand their supply functions of product:
senate-s In the mired consumer firm s; stem the integrability conditions thh Klc
roaming to iiotelhng 5 line integral. dead loss, and price potent: t1 (equal to the Baronc
"' function) are not satisfied l\or u hen it came to interpretation would it matter for
the salidity of the Pareto Iiarone Lerner conditions if they were While Hoteihng
gives separate COI'ISldCl'aUOn
to consumers uhen discussing excuses. the two treatments
. never adequately integrated
9. The last writer to be mentioned is Professor A. Bergson.20
He is the rst who understands the contributions of all previous
contributors, and who is able to form a synthesis of them. In
addition, he is the rst to develop explicitly the notion of an ordinal
social welfare function in terms of which all the various schools of
thought can be interpreted, and in terms of which they for the rst
time assume signicance. In view of his own very generous ac
knowledgments of the work of others, even where he himself had
independently rediscovered many basic theorems in an improved
form, it is regrettable that his contribution has received so little
notice. No doubt this stems in part from the mathematical
character of his exposition, and to the fact that he uses the rather
difcult notation of differentials throughout. The analysis that
follows is simply an enlargement and development of his impor
tant work.21
THE SOCIAL WELFARE FUNCTION
It is fashionable for the modern economist to insist that ethical
value judgments have no place in scientic analysis. Professor
Robbins in particular has insisted upon this point,"2and today it is
customary to make a distinction between the pure analysis of
3"A. Bergson, A Reformulation of Certain Aspects of Welfare Economics,
Quarterly Journal of Economics, LII (1938), 310334.
" In recent years Kaldor and Hicks have given an exposition of certain aspects of
welfare economics. N. Kaldor, Welfare Propositions in Economics," EconomicJournal,
XLIX (1939), 549552. J. R. Hicks, "Foundations of Welfare Economics," Economic
Journal, XLIX (1939), 696712. Mention should also be made of an important article
which indicates the modication in the Pigouvian analysis necessitated by the considera
tions of monopolistic competition. R. F. Kahn, "Some Notes on Ideal Output,"
Economic Journal, XLV (1935), 135. A convenient compact summary of welfare
economics is provided by O. Lange, The Foundations of Welfare Economics," Eco
nometrica, X (1942), 215228. An advance in the discussion is represented by T.
Scitovsky, A Note on Welfare Propositions in Economics," Reviewof Economic Studies,
IX (19-11),77-88. Because discussions of Free Trade illuminate the beliefs of economists
upon these matters, and because this subject provides a convenient illustration, it
would be desirable to review its literature. However, reference can only be made here
to the survey in J. Viner, Studies in the Theoryof International Trade (New York: Harper,
1937); to T. Scitovsky, A Reconsideration of the Theory of Tariffs," Reviewof Eco
nomic Studies, IX (1942), 89-110; P. A. Samuelson, Welfare Economics and Inter
national Trade," American Economic Review,XXVI II (1938), 261266;The Gains from
International Trade," Canadian Journal of Economics and Political Science, V (1939),
195205.
L. Robbins, An Essay on the Nature and Signicance of EconomicScience (London,
1932).
220 I-OUNDATIONS OF ECONOMIC Ah!/[LLS]?
Robbms qua economist and lus propaganda. condemnatzons. and
pohey recommendattons qua exttzen In practice, If pushed to ex
tremes, thls some hat schzzophremc rule becomes difcult to ad
here to and It leads to rather tedious crrcumloeutrons But m
essence Robbins :s undoubtedly correct 15th thinking nsa
pouerful deterrent of good analysrs and descnptmn, and etlucal
conclusrons cannot be dem ed m the same ma} that screntzc
hypotheses are Inferred or xerted
But xt lS not \ahd to conclude from this that there 13no room
m econom1csfor n hat goes under the name of n elfare economics
It ISa legmmate exermse of econonnc analys1s to examine the con
sequences of vanous \alue Judgments. nhether or not they are
shared by the theonst Just as the study of comparatne ethres IS
1tself a scuence 11Lean} other branch of anthropology If nt ts
approprrate for the econonust to analyze the \\ay Robinson Crusoe
durects production so as to maximize lus (curious) preferences, the
economist does not thereby commit himself to those tastes or m
quire eoncemmg the manner m Much the} ere or ought to hate
been formed No more does the astronomer who enuncxates the
prmcrple that the paths of phncts are such as to nnmmlze certam
Integrals care uhether or not these should be mmmuzed ne1thcr
for all e Lnon do the stars care Our about historical revue
should show that there )5 meat) and eighty content to the eld
of \xelfare econonucs \uthout mxclung next methods m economic
thought In saymg thxs I do not mean to unply that the eld ol
welfare economics has sczcntlc content because a number of xts
theorems do not requrre mter personal comparisons of utllt) , thts
after all us a mere detall That part Much docs m\ol\e mter
personal comp Insons of utlht} also has re ll content and Interest
for the sczentncanal) st exon though the screntzst does not con
Stder nt any part of lus task to deduce or \enf} (except on the
antbmmlegrmal Je;9.1)the value gudgments Mm .tmpbcatms he
grmds out In the same nay, the nnthematacal theory of prob
abrht} accepts as a prumtn e undened assumptzon hose mlldlt)
xsnot ntsconcern the m1tnl specucatlon of equally likely m cnts
the measure of \arlous classes or the collectn e. and 1t then
proceeds to grmd out the mathematlcal lmphcatlons of these and
subsrdnary hypotheses It xsonly [arr to pomt out, hoxxexer. that
the theorems enunerated under the heading of \\ clfarc economics
WELFARE ECONOMICS 22I
are not meaningful propositions or hypotheses in the technical
sense. For they represent the deductive implications of assump
tions which are not themselves meaningful refutable hypotheses
about reality.
Without inquiring into its origins, we take as a starting point
for our discussion a function of all the economic magnitudes of a
system which is supposed to characterize some ethical beliefthat
of a benevolent despot, or a complete egotist, or all men of good
will," a misanthrope, the state, race, or group mind, God, etc. Any
possible opinion is admissible, including my own, although it is best
in the rst instance, in view of human frailty where ones own
beliefs are involved, to omit the latter. We only require that the
belief be such as to admit of an unequivocal answer as to whether
one conguration of the economic system is better" or worse"
than any other or indifferent, and that these relationships are
transitive; i.e., A better than B, B better than C, implies A better
than C, etc. The function need only be ordinally dened, and it
may or may not be convenient to work with (any) one cardinal
index or indicator. There is no need to assume any particular
curvature of the loci (in hyper-space) of indierence of this func
tion. Utilizing one out of an innity of possible indicators or
cardinal indices, we may write this function in the form
W = W(zl, zz, - - -), (11)
where the zs represent all possible variables, many of them non

cconomic in character.
Between these zs there will be a number of technological
relations limiting our freedom to vary the zsindependently. Just
what the content of these technological relations will be depends
upon the level of abstraction at which the specier of the value
judgments wishes to work. If he is an out and out Utopian he may
wish to ignore various institutional relations regardless of their
empirical importance; indeed, he may go all out and repeal the laws
of conservation of energy and widen greatly the technological
productivities of the system. On the other hand, he may wish to
take as xed and immutable all social and economic institutions
except those relating to the Central Bank. (Indeed those of fatal
istic temperament may regard the restraints to be so numerous as
to leave no problem of choice.) In other words, the auxiliary
222 FOU\D ITIOhS OI" I COSOlNC [lb U ISIS
constraints Imposed upon the \1rnhles we not themethusthe
proper subject nutter oi uelhre cconmmcshut must ht when as
m ens
Subject to these construnts Much may he \mtten tht most
gener'zl u 1y ts
K (! :: ) ' 0 (12)
tere ml! prLsunnhly be m upper bound to ll (u on though no
umque nlue of the : s need correspond to thus urmmum leul)
If cerhm summum-. of reguhrlt) 'un. m uh It ouh! he possible
to tndtmte forum! conditions for the unumum tnmlung La
gnngnn multmhers untrlres r-mk 1nd qu: lrtttc forms (ltmu
under constnmt etc lloueur than : no mrtzeuhr pomt at
de\ eloplng this fornnllsm
Tht subject coull U1I mth these lmmhtws (re Il not for the
het thtt numerous Inthwhnls land ll of Interest to speculum
the form of " the mture of the \"Irnhles : uul the mum of
the constnmts
(l) l or one thing paces 1n not usually Included In the utlhre
function use ucept ur) :mlueetl) through the Lfftctsof dzftr
ent puces 'tnd thus upon the qu umtusof consumption uorL etc
(2) Also cut un of the \ nrnhlu cm lx. thought of 1s referring
to '1mrtteular md ultnl or luml} e l. one of the : 5 m1} he the
amount of tn consumed In; John ]onu or the amount of unskilled
hbor Inch he prox:des
(3) It 15often further rimmed tlnt the qu'mtmes of 't gncn
commodity consumed by one Indmdlnl 1re of the same type 15
those consumed h) mother technical!) tim me'ms that cert'un
of the \ 1rnbles enter m the technical sidecontinuonsm sums u Inch
rchte the tot'tl 'unount produced of '1commodlt} to Input regard
lessof the ultmntc distribution of tlnt output \\ lntm er rm) be
soul 1bout the 1(hmssnlnht) of tlus the matter ts St!" worse hen
1 south! 1ssumptton IS mule concerning the homoeenczt) of the
\1noussen1ces prowlcd by differentmtlmdmls llowmer curt
nf m the most morons sense mch tndmdml s talents nrc umquc
soexety nrely 1115the tune or mtlence to learn to 'tpprccmtc the
uxor of each mm and m the 1bsence of perfect screening 0f
dn'crent 1ndmdu1ls 1t does treat them 15 1f they en. perfectly
substttutablc thus for our purposes the) may be assumed to bc so
n many cases Tlus does not mem tlnt wc work With a smgle
grade of labor; on the contrary, the number of grades may be very
large and the classication minute, but it is assumed that there
are many individuals in each grade, actually or potentially.
(4) Not infrequently it is assumed that a given grade of pro
ductive service may be used indifferently in a number of uses.
Technically this means that certain of the 2's enter into the welfare
function only as certain sums. As Bergson has pointed out, this
involves implicit value judgments, so that Robbins when he dis
cusses the problem of allocating resources as between alternative
uses so as to maximize (in some sense) output or personal utility
is not able even on the Robinson Crusoe level to avoid these no
tions; or rather, if resources are not assumed indifferent between
at least two uses, few interesting marginal conditions can be
deduced.
(5) A more extreme assumption, which stems from the indi
vidualist philosophy of modern Western Civilization, states that
individuals preferences are to count. If any movement leaves
an individual on the same indifference curve, then the social welfare
function is unchanged, and similarly for an increase or decrease.
Actually, an examination of the principles of jurisprudence, the
folkways and mores, shows that in its extreme form this assumption
is rarely seriously proposed. Even sane" adults are not permitted
to eat and drink what they think best, individuals cannot sell
themselves in order to consume more in the present, milk ration
tickets cannot be exchanged for beer at the will of the owner, etc.
But economists in the orthodox tradition have tended to con
sider the above cases as exceptions.23 However, in recent years
many economists, Frank Knight being a notable example, have
2Consider, however, the following interesting quotation from Edwin Cannan who
was very much in the classical tradition: We shall never decide whether to put a penny
on beer or to further steepen the supertaxon incomes by considering how much the
loss of a penny pinches the beer-drinker and the duke: we shall, and we do, decide lt
by making some rough estimate of the aggregate advantage in the long run of the two
methods to society at large. For example, if we nd that cheaper beer means better
food for underfed children while less super-tax means more training of horses to run fast
for a short distance with a very light burden, we incline to the super-tax: but if we nd
cheaper beer means more beer for drunkards and less super-tax means more houses for
the people to inhabit in comfort and health, we incline to the beer tax." From an
Economic Journal review of Sir Josiah Stamps Fundamental Principle: of Taxatwn 171
the Light of Modern Developments, reprinted in An Economist's Protest (London, 1927),
p. 279.
224 FOUNDATIOAb OP ECONOMIC ANALYSIS
msxsted upon the degree to whnch mdmdual tastes and wants are
qocrally condmoned by adxertlsnng and custom so that they car
hardly be smd to belong to hun m any ultlmate sense All this ;:
recogmzed m the u 1ttlczsmof the soap box speaker n ho satd to the
recalcutrant hstcner, When the rexolutlon comes you wnll eat
straubemes and cream, and hke lt' Attention should also bc
{alled to the (act that even the classrcal economlst does not laterally
have the mdmdual mind, so much as the family. of course som:
hardy souls wnll pursue the mll o' the wxsp of sovereignty unthtr
the family so as to reduce even these collectnve mdrerencecurve:
to an 1nd1vnduahst1cbasrs
(6) One does not haxe to be a John Donne then to nd fault
n 1th the above assumption, especzally ll e consrdcr the closel)
related assumptxon that an nntlmdual s preference depends on!)
upon the tlnngs \Vlllcllhe consumes and not upon what other:
consume As Veblen charactenstncally poxnted out. much of th:
motxvatlon for consumption lSrelated to the fact that others do 01
do not have the same tlung Conspicuous expendlturc, keepxng
up w1th the Joneses," snob appeal, maintenance of face, are tm
portant m any reahstlc appraisal of consumption habits, and If me
turn to the eld of pouer analysrs, xt ISnot only on the natuona
scale that satlsfactlons" are relatlve and dog 1n~themanger tacuc:
are ratxonal
If thls smth assumptton IS not made, many of the conclusxon:
of welfare economxcsml! remam \ahd, but they ml! require mod:
catlons to allow for certaln external consumptnon econorme:
not dISSImllaranalyncally to the external technologlcal econonue:
md dtseconomles of the Marshall Plgou type
(7) All of the above assumptmns are more or less tacxtly ac
cepted by extremely dwergent schools of thought The next as
sumptnon Involves a more controverszal \alue Judgment, but om
whlch has been characterlstrc of much of modern thought of tht
last century, and WlllCllIS especnally typical of the beliefs of the
classrcal and neo classncaleconomists It :: that the welfarefunctzon
:s completely (or very nearly) symmemcal wzth respect to the con
sumptwn of all mdunduals
Taken 1n connectlon w1th prevxous assumptions, m ltS strict
form 1t ISnot cons1stent w1th the patent fact of consrderable dlffer
ences mdmduals' overt preference patterns Thus. m addrtlon
to involving a very signicant value judgment, it also involves a
very denite assumption of fact. This was not appreciated by
economists, who tended to believe in the desirability of an equality
of income, leaving it to the individual to determine the exact form
of his consumption. However, it is easy to show that the rule of
equality of income (measured in dollars, numeraire, abstract pur
chasing power) applied to individuals of different tastes, but made
to hold in all circumstances, is actually inconsistent with any de
terminate, denite W function. Equality becomes a fetish or
Shibboleth, albeit a useful one, in that the means becomes the end,
and the letter of the law takes precedence over the spirit.
For the decision of equal incomes as optimal in one situation
implies a certain relative well-being as between vegetarians and
non-vegetarians; at different relative prices between vegetables
and non-vegetables an equal distribution of income can no longer
be optimal. Actually this does not render invalid the reasoning
based upon this seventh premise since the adherents of this view
implicitly held that individuals were very much alike, and if given
equality of treatment would develop the same want patterns.
Moreover, they could in all logic take the milder position that a
great deal less inequality than existed in real life would be desirable
even if one did not believe in complete equality.24
In a similar way, the belief that the individual should rightfully
receive his imputed productivities is not consistent with a W func
tion having properties (1) through (6). A change in the techno
logicalsituation will alter individuals fortunes so that the nal result
cannot be optimal if the initial situation was deemed so. Perhaps
the bourgeois penchant for laissez-faire is the only case on record
where a substantial number of individuals have made idols of
partial derivatives, i.e., imputed marginal productivities.
One could similarly multiply expressions and beliefs which are
part of everyday parlance, which upon examination turn out to be
inconsistent and meaningless. The slogan, the greatest good for
the greatest number," was shown by Edgeworth to be one 511611;
It is not the purport of the above lines that the use of a welfare function leads to
a belief in inequality as compared to equality. It merely shows that equality Ofmoney
income where there is diversity of tastes involves the equality Of "Othmg important.
lt is in lesser degree like the Anatole France aphorism concerning the equality Of the
law in its treatment of rich and poor. Before Bergson's treatment one could ..
sensed, but not analyzed denitively, this subtlety.
and we might add the dictum each to count for one and only one
As Professor nght has ceaselessly msrsted Western Man ISa
hodgepodge of behefs stemmmg from dlverse and mconmstent
sources Fortunately hlS hfe lS suiCIently compartmentahzed so
that he can assume hls vanous roles w1th a tolerable amount of
ambtgmty m each and only the most mtrospectxve worry about
th1s enough to become dlsorgamzed
(8) A nal essentmlly unnecessary assumptlon whtch was espe
elally characteristtc of the last generation of economnsts was the
demtwn of the welfare functwn whzch was to be maxzmzzed as the
sum of cardmal utzlmes experzenced by each mdwzdual Before the
t1meof Bergson this was not uncommonly met even m the advanced
llterature and vestlges can be found today It stemmed from the
mam Utllltarlan stream of economic thought when ut111tywas used
mterchangeably m a behav1orlst1c m '1 psychologlcal 1n a physno
loglcal In an ethical sense
It was not uncommon for older wrlters to ponder over the
questlon as to whether ut111tywas bemg maxlmlzed or whether
pam was bemg rmmmlzed whether by and large man was operating
m the red below absolute zero but makmg the best of a bad lot
The answer depended upon the author s theology and endocnne
glands at the moment Paley Sldgwmk and others could sen
ously ask whether 1t was better to have a tremendous population
each contnbutmg a llttle to a vast amount of Socnal Utxhty or
whether 1t was better to have less Soc1al Utlllty provxded Its
average amount per head v.as maumxzed
In the eld of Public Fmance the assumption of addntwe 1nd:
Vldual utlhtnes plus the law of dtmtmshlng ut111ty were used to
Justlfy the xmposxtlonof progresswe taxes In 1ts most swecpmg
form this doctnne set up m1mmumaggregate sacnce or maxunum
total ut111tyas approprlate goals of actlon Thts goal can be ob
tamed only If the marginal ut111tyof Income (after taxes) 15equal
for all mdwxduals or 11'1nd1v1dualsare essentially alike only w1th
equal Incomes for all 25 On the other hand the cnterxon that a
gwen sum of taxes should be raxsed so as to lead to equal sacnce
for all IS a much more conservatlve doctrine followmg 1t we can
"F Y Edgewonh The Pure Theory of Taxat on Economw Journal V
897) 550-571
" uurn1\u uuuuuuu L.) 227
only be sure that taxes should increase with income, but not
necessarily in proportion with income.26
Today such arguments are not very fashionable since it is as
easy to assume ones conclusions with respect to appropriate policy
as to assume the premises of these arguments. Not only is the
former more direct, but it is more honest. Nevertheless, some of
the considerations which enter into the above arguments are latent
in much of modern discussion and thought. In the distribution
of war burden the moderately well-to-do tend to point to their
rather large sacrices and to call upon the lower classes to share in
these and make new ones. The relatively poorer farmers and
laborers concentrate upon how much the rich have left after they
have made substantial sacrices, and how little they themselves
have in any case, not in comparison with what they had before the
war, but compared to what they consider fair. Whether war
times are the appropriate times to redress ancient wrongs or not,
it is too much to expect that the bargaining advantages which the
war brings will not be used for this purpose.
In connection with this eighth assumption it was implicitly
assumed that real income could be treated as a homogeneousquan
tity to be distributed among individuals. This could be literally
true only in a one-commodity world, or in a world where all relative
prices were xed constants. Actually prices will vary depending
how money income is distributed. Strictly speaking, therefore,
the real judgments embodied in the welfare function must be judg
ments concerning a multitude of diverse goods. This would be
The condition of equality of sacrice is satised at each level of income if
U(X) _ U(X t) = constant for all X.
Differentiating this so as to determine the explicit change in t with respect to X ,
dt _ U'(X z) U'(X)_
dX _ U'(X :)
Because of diminishing marginal utility this is positive. but if we wish to have pi:
grCSSVC taxation, the elasticity of income after taxes against income before taxes must e
less than one. But
X d(Xt) _ XU'(X)
Xz dX _ (Xt)U'(X-t)
which is less than one if, and only if, the elasticity of the marginal utility curve is 155
than unity. Thus, for Bernoulli's law of utility, equality of sacrice would imply pro- _
,..
portional rather than progresive taxation.
quite a problem eVen for a man with denite opinions and great
preoccupation with value judgments If, however, he takes refuge
in assumption (5) that individual preferences are to count, the
individual can be left to decide for himself how he will spend his
money at given prices. Our ethical observer need only decide
then what his preferences are as between the given levels of satis
faction of different individuals
It might be thought that our ethical observer, even if the indi
\.iduals themselves had no unique cardinal indexes of utility, would
have to nd cardinal indicators But this would be quite incorrect
Of course, if utilities are to be added, one would have to catch hold
of them rst, but there is no need to add utilities The cardinal
utilities enter into the W function as independent variables if as
sumption (5) is made But the W function is itself only ordinally
determinable so that there are an innity of equally good indicators
of it which n be used. Thus, if one of these is written as
W = F(U. U. ). (13)
and if we were to change from one set of cardinal indexes of in
dividual utility to another set (V, l, ), we should simply
change the form of the function F so as to leave all social decisions
invariant Thus, let us move from one conguration of goods
going to the different individuals to another conguration which
leaves W unchanged or which xs socially indifferent. Then no
redenition of the Us or of F can change this fact the social in
differenceloci are independent of cardinal numbering And in this
terminology the signicance of assumption (5) that individual
tastes shall count is contained in the requirement that the social
indifference slopes between two goods gomg to the same mdmdual
are exactly the same as the individuals indifference ratios As
sumption (6) adds that these curves are unaffected by changes in
the goods going to other individuals
" Even If one Wishedto add ut1lnt1es,rtwould seem silly from any ethical wen pomt
to have ones opuuons as to correct taxation Inuenced by how consumers spend their
income on goods Yet that Is what the recent attempts to measure marginal utility
must amplyll' they have any pretensions to relevance for policy R Fnsch.New limited!
of Measuring Marginal Uhluy (Tbmgen, 1932) I Fusher, A Statistical Method for
Measuring'Margmal Uuhty' and Testing the justice of a Progressive Income Tax."
Economic Essays Contribzdrd m Honor of John Bates Clark (New York, 1927)
MATHEMATICAL ANALYSIS
The assumptions discussed above can be given mathematical

formulation. Thus, assumptions (2) and (3) imply that the impor
tant variables of our system can be regarded as a number of com
modities and productive services (X 1, X 2, - - ,X,,, V1, V2, . . ., V).
These totals can be distributed among s individuals of the system;
the amounts going to each'can be written with a superscript which
identiesthe individual and a subscript identifying the commodity
or service. (While productive services could be written as negative
commodities, I have chosen to conform to the more common pro
cedure encountered in the literature.)
x.=ix.-k, (i=1,---.n)
k=1 (14)
Vi: vjk' (j=11"'77n)

k=l
The social welfare function involves only the amounts going to
each individual, not prices or totals. Therefore, equation (11)
is specialized to
IV= l/l/(xll' ...,xul; . . .;xl, . . .xn3;
v11. M; ---:W. -~,v..'). (15)
This exhausts the implications of the rst four assumptions.
Assumptions (5) and (6) further specialize (11) so that the
welfare function can be written as
, : /VEU1(xll, . . ., xnl;'011, , __, ml), _ __,
Un(xla' ' ' ': xni vla: ' ' ', wmv]. (16)
where the cardinal forms of the Us and the W. are arbitrary.

ASSumption (7) is meaningless unless the respective U's can be
made identical; but if this can be done, W must then be a sym
metric function of the Us. Assumption (8) requires that .there
exist a cardinal W and cardinal Us such that W is a surnmatlonpf
the U'S On the anthropological level this involves (aside-from its
arbitrariness) denite restrictions on the social rates of indifference
between the commodities and services offthe same and different
individuals. These are similar to the empirical restrictions of
independent individual utility as discussed in chapter vu.
In the prevxous sectnon 1t was noted that the techmcal re
stramts must be assumed n 1ththe same arbitrarmess as the welfare
functron Itself However, smce the formulatxon of general equr
hbrrum by Walras rt has been customary to take as gwen by the
engmeer the fundamental relatxonshnpsbet een mputs and outputs
and that productlon Itself takes place rms or Industries whlch
are distinct from the mdlwduals havmg no value m and of them
selves Under modern 1ndustrral conditions thus ISnot unreahstnc
But even here many mterestmg altematwes can anse What one
calls econormcs economic engmeenng, engineering, etc , rs to a
consnderable degree a matter of chouce One can assume that all
production dec151onsmvolvmg relatrve marginal productwntlcs are
the concern of the engmeer, or economic engmeer, and that the
economistcan take as already derived a transformation relationship
between the X s and V 5 of the form
T(Xl X2' : Xi V1 V2, ) Vn) "= 0 (17)
Thxs xmplmt relation 15Interpreted to gwe the maxrmum amount
of any one output as of gwen amounts of all mputs and all other
outputs or the minimum amount of any one Input as of gwen
amounts of all outputs and all other mputs
But 1fone starts from more prrmztlve technological assumptions
such as the production functions of each commodlty, the trans
formation locus ISa derwcd theorem and not an axiom Behind
rt he many Interestmg optlmal production conditions mvolvmg
marginal productlvrtles and other magnttudes usually thought of
as economic rather than engmeenng
PRODUCTION CONDITIONS
In a synthesrs of welfare economrcs ne can derrvc rst a set of
production conditions which rcqmre for thexr valrdnty the u eakest
of all ethlcal assumptrons szmply that more of any one output, other
commadztzes or sermces bemg constant. z: dcszrable, smnlarly, les:
mput for the same output: 15deszrable
Under the srmplest technological conditions ne can take as
gnen the productron function relatmg each output to the mputs
devoted to 1t
where the rst subscript indicates the kind of productive factor
and the second subscript the commodity to which it is applied. Of
course, the total applied to all commodities is given by
Vj=vj1+vj2+m+vjn~ (j=l.~-,m) (19)
Ordinarily it is assumed that the production function possesses
partial derivatives (marginal productivities), but the above form
ulation includes the case of so-called xed proportions or xed
coefcients where the production function has contour lines with
corners.
Equations (18) and (19) represent (n + m) relations. If we
specify arbitrarily all but one output (service), we can maximize
(minimize) the remaining one. This is an extremum problem in
which there are auxiliary constraints. The rst order partial de
rivative conditions can be expressed directly as proportionalities
and rank properties of the matrix of rst partial derivatives of our
functions. But it is illuminating to express these directly by
means of the artice of Lagrangean undetermined multipliers. To
do this we set up the function
=7F1X1+7F2X2+
+7Tan+>\1V1+>\2V2+ '" +Xme (20)
and pretend to maximize it treating all the variables as if they were
independent, and treating the 7rsand Asas (undetermined) con
stants. If the correct secondary extremum conditions are written
out in full, it will be seen that they are quite different from those
that would have to hold if the @function were really to be maxim
-ized. Only by accident would these secondary conditions coincide;
only by accident would (I)not be at a minimum or would it be at
an extremum rather than simply at a stationary value.
This may seem to be an esoteric and recondite point of little
practical signicance. It is the primary contention of the present
work that everythinginteresting is contained in the inequalities as
sociated with an extremum position, rather than in the equalities.
This is no less true of the eld of welfare economics. From a
deeper point of view market prices, regarded as parameters by perfect
competitors, are nothing more nor less than Lagrangean multipliers.
The Langrangean expression <13 corresponds to Barone's <19
function
or to Hotellings price potential and can be regarded as the value
432 I'OUNDATIONS 01 ECONOMIC ANALLS'IS
of output or national product, expressed m terms of money or any
numeralre, and generahzed by the subtraction of factor costs
1f the game of competmon leads to optimal conditions It does
so partly by accxdent For nt xs prccxsely under the cond1t1ons
favorable to the mamtenance of atomlstlc competition (briey
absence of decreasmg costs) that the secondary condxtnonsof the
correct mammum problem agree w1th those guaranteemg the max
Imlzatton of the money value of output as of xed prices Where
there are substantnal technologlcal moreasmg returns, competztton
as an empmcal phenomenon breaks down It IS1nsuch condxtlons
that collecthsm IS hkely to be considered sertously as a socxal
pollcy If then a socmllst reglme should mast upon mechanically
playmg the game of competxtxonthh prlces regarded as parameters
a la Lange," Its managers would rush away from the mlmmlzatxon
of @, even though thus mmzmzzatzon may be precuscly what 15
rcqmred under decreasmg cost 1fwelfare rs to be maxzmzzed
The moral 15not that pnces should cease to be related to mar
ginal cost or that plannlng under soc1ahsm IS 1mposs1ble It IS
snmplythat the decentrahzed operators m a planned sootety should
refram from a hteral apmg of atomxstlc, passwe parametrlc prxce
behaxlof Instead of pretending that demand curves are mmtely
elastic when they are not the correct shape of the curve 18to be
taken Into account 'lhls does not mean that the decentrahzed
operators should take account of the1r mfluence on price as a
monopohst would
After the ehmmatxon of Lagrangean multnphers the rst order
maumum condttlons take the form
a_X.
69]: _ _ vm; _ TX; _
E -
EE r (ale, l 123, (21)
av: aime
" O Lange amcle On the Economu: Theory of Sacmhsm B E L pp ncott ed

(\lmneapohs University of Minnesota Press 1938) PP 55142
When we come to the full statement of welfare condmOns ll. be seen that
unusual d fcultaesdo anse m the decreasmg cost case m determining whether a gwen
maximum posmon represents a manmmn maxmrorum or whether the number of differ
; ated products should he reduced
or the equivalent form
a av, Ti" .
a1=--o=a)n=7- (],r=1,---,m) (22)
Ov,-1 avr)!
In words this takes the form: productive factors are correctly

allocated if the marginal productivity of a given factor in one line is
to the marginal productivity of the same factor in a second line as the
marginal productivity of any other factor in therst line is to its mar
ginal productivity in the second line. The value of the commonfactor
of proportionality can be shown to be equal to the marginal cost of the
rst good in terms of the (displaced amount of the) second good.30
Geometrically these conditions can be easily derived in the two
good, two factor case by means of a Jevons-Edgeworth-Bowley
Lerner diagram consisting of a box whose respective sides are
equal to the available total amounts of the two factors. Any
point within the box, if oriented with respect to the lower left-hand
corner, can be thought of as representing the amounts of the two
factors used in the production of the rst good, and the iso-product
contour lines can be sketched in. The same point when referred
to the upper right-hand corner represents the allocation to the
second good, whose contour lines can also be superimposed on the
same diagram.
If we specify the amount of one of the goods and restrict move
ments to one of these contour lines, the optimum position is reached
only when we have touched the highest output line of the other
good, or at a point of tangency of the opposing iso-product lines.
The geometrical diagram indicates the correct secondary condi
tions which are quite different from conventional diminishing re
turns. The locus of all such points of tangency, which can appro
priately be called a generalized contract curve, represents the
innity of optimum positions. If along this locus we read off the
amounts of the respective products, and plot one of these magni
By introducing as variables inputs and outputs of different dates optimal behavior
over time can be included in the above formulation. However, when this is done, it
will be seen that contrary to the belief of most economists since the time of Bhm
Bawerk, no single real interest rate is implied for a capitalist or socialist state.
Equality would be a necessity only in the highly unusual case where relative prices of all
goods remain the same over time.
tudes against the other, the resulting locus is the substitution,
transformation. or opportunity cost eure, T. The slope of this
curve at any pomt represents the marginal cost of one of the goods
in terms of the other, or the ratio of marginal costs expressed in
terms of some third magnitude.31
The substitution curve is drawn up as of given amounts of the
factors of production and will shift with any changes in the latter.
For a given transformation curve to be of any relevance, the factors
of production must be regarded as being indifferent as between
different uses
If constant returns to scale and only one factor (perhaps equal
to a composite dose) are assumed. we have the classical case of a
straight line transformation curve if the first of these assump
tions is made. but there is more than one factor of production,
unless the goods should Just happen to use the factors in the same
proportion, the transformation curve will be concave to the origin
because of the law of diminishing returns to varying proportions.
The curve can also assume this shape for other reasons relating to
returns to scale. In the decreasing cost case. however this arises,
the curve Willbe convex to the origin.
In all of these cases let the firms or planners really take prices
as gwen and attempt to maximize value of output or tb. In the
constant returns case there Willbe complete specialization on one
of the goods (the one with comparative advantage) or complete
indifferenceat the critical price ratio equal to the cost transforma
tion ratio If both goods are consumed, and if the economy is a
closed one, the latter critical price ratio is sure to be the only
relevant one.
In the increasing cost case at given price ratios the rm will
For the geometrical picture see W F Stolper and P A Samuelson, Protection
and Real Wages," Rem-w of Economic Studies, IX (1941), 58-44.
It is only when the last two unnecessary and unrealistic assumptions are made
that the socalledopportunity cost doctrine is even formally valid Even under these
conditions the usual formulation is rather a muniho )umbo of high sounding gibberish,
which does not state the conditions of equilibrium in a very direct way All this is
further complicated by the fact that most enunCiationsof the opportunity cost doctrine
are purely verbal leisure is treated as a displaced good! IneVitably, therefore, when
the Opportunity cost doctrine is carefully stated and qualified, it degenerates into the
full conditions ol general equilibrium in which factor supply and preference equations
must he introduced if only as inequalities This is not to imply that one must accept
dubimis psychological language and interpretations of classical real cost theorISts
come into equilibrium at the levels of output which make prices
proportional to marginal (transformation or factor) costs. How
ever, in the decreasing cost case the planners in a socialist state at
tempting to maximize at xed prices would concentrate on one or the
other of the two commodities. But this would be optimal only in
an open economy which could really trade with the outside world
(in unlimited amounts) at the given price ratio. In a closedplanned
state it might be desirable to produce something of both, to have
price ratios equal to marginal cost ratios even if that meant a mini
mization of CI). If we regard the passive approach of the decen
tralized planners as an equilibrium process, it is clear that the cor
rect optimum, being a minimum and not a maximum, would not be
an equilibrium point at all. Or if the planners acted like Balaam's
ass, it could be regarded as a highly unstable equilibrium point
from which they would rush at the slightest disturbance.
Thus it is only accidentallythe rst two casesthat com
petition is optimal. For it is in those cases that the Lagrangean
expression (20) is at a maximum when we are at the correct maxi
mum of (18). In these accidental cases, if we like, we may think
of our production decisions as being independent of preferences, in
the sense that they are uniformly best with respect to the most
extreme weightings of the commodities. Under increasing returns
we may even never settle on societys best transformation curve.
Actually our production conditions might have been introduced
in more complex form than as single, isolated production functions
of the form (18). It is to a certain extent arbitrary as to how much
preliminary maximizing behavior we consider as being done by the
engineer and how much by the economist. Under competitive
conditions the relevance of prices to the decision is taken as t.e
dividing line; but for welfare purposes we may dispense with prices
completely. However, it would seem somewhat awkward then to
term the problem one of engineering; still some economists
might prefer to begin with the transformation function, T.
If we examine production possibilities more closely, it will be
apparent that in many cases it will not be possible to end up With
a single transformation function, T. Thus suppose that commodi
ties divide up into two or more groups such that no productive
factor is used in more than one group. Then there will be at
as many independent transformation curves as there are '
Of course, formally these can be combined into a single implicit
equation, e.g., by equating their summed squares to zero, but this
is trivial. No indeterminacy arises, however, from the multi
plicity of transformation curves; we have fewer choices to make
and therefore fewer marginal conditions.
PURE EXCHANGE CONDITIONS
The production conditions of the previous section were deriv
able on the basis of rather mild assumptions by virtue of the fact
that more of all goods, however they are to be divided, would seem
better than less. These conditions are still only necessary. They
are not sucicnt. since decisions must still be made as to how to
divide up given available goods, and as to which of the available
amounts of production will actually be used.
Even if the last of these two decisions is made, there remains
the problem of dividing up a given total of all commodities and
services among individuals. Once there has been specied a clearly
dened W function, the nal Optimum point can be easily deter
mined. But at this point after assuming, if only implicitly, the
rst six serious assumptions, the modern welfare economist be
comes timid and hesitates to make assumption (7) or indeed any
other specic welfare function. He has been conditioned against
making ethical assumptions. If possible, therefore, he would like
to develop further optimum conditions which are within broad
limits independent of the form of the W specied. In fact, as has
been known since the time of Pareto at least, it is possible to
gratify this wish and to specify still another set of necessary,but
not suicient,conditions which must prevail as between individuals.
Here the simple assumption is made that W is undened, or
unspecied, except to the extent that it is a monotonic increasing
function of the individual U's. We are not able to plot in the
(U, U)|plane a locus of points along which W is constant. We
know only that a movement in the. rorthcast direction altheplane
increases W and is good, a movement in the southwest direction
decreases W and is bad; a movement into either of the other two
quadrants taken around a given point is indeterminate.
In words, the simple assumption is made that only if all indi
viduals are made better (worse) off can we denitely state that a
movement is good (bad). Otherwise we must reserve judg'
ment. Mathematically it is clear that for the levels of utility of
all but one individual being arbitrarily specied, it is a necessary
condition of equilibrium that the utility of the remaining individual
be at a maximum, subject to the condition that there be xedtotals
of all goods. For
i x! = Xi, }: 2:,1= 17,, (23)

J=1 i=1
Ui(x1i, 133',in", vm) = } (J' # k)

we must maximize
U]: = UL-(xlk . . . , xnk pli-, . . ., 21"."). (24)
By using Lagrangean multipliers or by direct methods, it can

be easily shown that in equilibrium the ratio between marginal
utilities of two goods consumed by one individual must be the
same as the ratio of marginal utilities of the same goods to any
other individual who consumes the same goods. If one or more
of the goods is absent from the consumption of an individual, cer
tain inequalities can be introduced to generalize the above condi
tions. \Ve may write the rst order conditions of equilibrium in
the form
===l_f. (25)
U, (1,2 U;
It will be noted that only ratios of marginal utilities for the same
individuals are ever involved. Thus, there is no need for nu
merical utility even for a single individual and no need to compare
the utilities of different individuals.
Graphically the equilibrium may be represented by the same
box-type diagram as was mentioned in the last section. The di
mensions of the box represent the xed totals of the commodities
in existence. Any interior point oriented with respect to the
lower left-hand corner represents the amount of consumption of
the rst individual, and the contour lines are now to be interpreted
as indifference curves. Referred to the upper right-hand corner,
a point represents the second individual's consumption, and his
contour lines may be superimposed on the other. Specifying the
Equilibrium conditions regarding the factors of PdeUCOneed not be indicated
separately if it is remembered that they may be treated as negative commod
utxlxty of the second 1nd1v1dual,we maxxrmze that of the rst by
movmg along the gwen mdl'erence line of the second 1nd1v1dual
untrl ne reach (at a pomt of tangency) the hlghest 1nd1flerence
curve of the rst Because the ongmal specncatnonof the second
man's utxhty was arbitrary, the nal equrhbnum xsalso arbitrary
and not umque The locus of all such arbltrary pomts 15,of
course, the famrhar Edgeworthran contract curve and represents
the set of ponnts satrsfymg necessary exchange condmons
After the dtscussxon of the prewous section 1t Willnot be neces
sary to reiterate at length the contention that the equahty of the
xanous ratxos can be expressed thhout Lagrangean multrphers
and wrthout prlce ratlos, moreover, m by no means 1mp0551ble
cases, even playing the game of competmon would lead away from
rather than to ard the correct optimal posmon As before, thus
lS because the secondary condrtlons for a constrained maximum
are not the same as the secondary condxtxonsnecessary hterally to
brmg the Lagrangean expressron to an extremum
It Willbe noted that from any pomt off the contract curve there
exnsts a movement toward rt uhlch would be benecxal to both
Indrvrduals Thls 15not the same thrng as to say, wnth Edgeworth,
that exchange wl fact necessarily cease somewhere on the con
tract curve, for m many types of brlateraI monopoly a nal equr
hbrmm may be reached off the contract curve Nor 131t the same
as to say that pomts on the contract curve are better than pomts
off the contract curve Later I shall dlSCLISS the correct formula
tron of the sngmcance of thIS condltlon
But rst rt ISn ell to write down the full rst order condltlons
of equlhbnum wluch must hold 11'we are to have an optimum of
both productlon and exchange, that ls, If we combme w1th the
condmons of thlS sectnon the condttrons of the last sectlon Then
(1) ne must have a common margmal rate of zndzerence betweenany
mo goods for etery matvtdual, this common andz'erence 111110must.
moreozer, be equal to the ratzo at wind: one of these goods can be
transformed mto the other on a productzon sense. the transformatton
to come about as the result of transferring any resource from one
good 3 productzon to the others (2) We must have for all mdwzduals
a common who of mdzerence between supplymg more of any factOT
' praductzon and enjoymg more consumptzon of a gwen g00d: th
common ratio must be equal to the rate at which supplying more of
that factor results in greater production of the good in question.
Mathematically, as Lange has shown,
:} Un T_Jn
U3} = . . . = sz : Tx_ (ii,7.
: 1, . .., )
Uk1=___=_ULJ__7_ (i=1,---,1z) (26)
UZ,-1 U,: " am,, _ Tx (k = 1, . . ., m)
These conditions may be rewritten in a variety of ways but only

being (at the rst order level) necessary conditions rather than
sufcient conditions, they are necessarily less in number than the
number of the unknowns of the system. This is after taking ac
count of the fact that the total of factors used in every use and the
total of commodities consumed must equal the sums of these mag
nitudes over all individuals, i.e., even after taking into account
equations (14). We are still short (s 1) equations. These can
only be supplied on the basis of denite assumptions concerning
how different individuals enter into the W function. This will be
discussed in detail in the next section.
Returning to the combined production and exchangeconditions
of equations (26), I should like to point out various alternative
formulations of them. First, it is easy to derive from them the
conditions that the ratios of indifference between any two factors
must be uniform for all individuals, and equal to the relative
marginal productivities of the two factors in any line of production.
Second, it is sometimes convenient to interpret the rst of these
equilibrium conditions as specifying the equality of price and mar
ginal utility ratios to the ratz'osof marginal cost. There is occa
sionally some uncertainty in the literature as to whether prices
should egual marginal costs or just bear the same percentage dis
crepancy in every line. The distinction is especially important if
decreasing costs are very prevalent, for then it is impossible for
capitalistic monopolies or socialistic bureaus to recover their .full
costs of production if they charge only prices equal to marginal
costs. Recourse must be had to direct taxation to cover the differ
ence or to the imputed revenues accruing to state-owned factors Of
production, revenue which in a world of constant returns to scale
would be available for distribution to the citizens of the state.
So long as only pncc ratios are used 1t 15not necessary to state
the umts m Inch costs are measured dollars labor fertthzcr or
opportumty substatutnon costs In the usual exposxtlon partnl
equdlbnum conditions are 1mphc1tly assumed and costs \\ ages
and pnces are expressed m terms of dollars ln these condmons
must prices be equal to margmal costs or ]UStproportional> The
answer 16equalztyIl ne take mto account the conditions relatmg to
factors of production such as are embodied m the second part of
equatrons (26) lf all factors of productron crt mdr'crcnt bc
tneen dl'ercnt uses and completely xed m amount (the pure
Austnan case) then . could dispense mth these condttrons and
proportionality of pruccs and marginal cost uould be sufcnt
But If e drop these high!) SptCl'll assumptions for \\lltCllthere IS
not m any case empmcal or theoretscal \\.arrant then al'all puces
were prOportzonal to (say double) marginal costs ue should not
have an optimum sntuatlon By uorklng a little more or less
everybody could be made better off smce the descnbed Sttuatlon
the preferred terms at uhrch they e\change lezsure and goods are
not equal to the true productxvnty terms at value]: they can be
transformed :nto each other
I need not remmd the reader that rst order conditions are
really of secondary Importance as compared to the full unequahtles
unphed by an optimum posztlon Also It should not be necessary
to show m detail that the abox e conditions hold only If every factor
of productlon ls actually used m excry [me of productlon and Il
somethmg 15 produced of eyery good and If all production and
Indifferencefunct1onshate contmuous partial dem atn es If any
of these condmons fall the number of equalities may be dmnmshed
but every case the correct general Inequahtzcs mll preyent there
bemg any essential amlngmtzes the ChlIZlCthlZlthof the
Optlmum posxtton A closely related problem anses hen there
are multtple posxtlonsof ethbrmm \\ 1th strongly come\ m
dlfference elds and strongly concayc cost {uncttons such cannot
occur But there ISno reason why human nature should dlspla)
the ample regular propertles which the observer economist nds
convement And 1th the essence of the deereasmg cost Increasmg
returns lumptness 1ndwt51b1htyphenomena that unproper eurva
ture on the cost Sldc should enter the pICtllIc
Especrally where differentiation of product anses as a result of
monopolistic competition or as its cause, there arises the problem
as to whether or not a commodity should be produced at all. If it
is produced, the marginal cost conditions of (26) should be realized,
but there may be a better maximum where none is produced.
Here the extremum position is of the corner type, and the conven
tional equalities must be replaced by inequalities. This involves
decision making at a distance; we cannot feel our way to the op
timum, step by step, but must boldly experiment with diverse
combinations. Where such all or none phenomena are con
cerned, things often get worse before they get better, and so deci
sions im kleinen will not sufce. Even aside from the difculties of
the paragraph after next, we cannot decide that a thing should be
produced if a perfectly discriminating monopolist or government
bureau could recover its total cost by some sort of a take-it-or
leave-it oer. In these cases involving nite decisions we must
ask consumers (or Robinson Crusoe) whether a given abundance of
fewer commodities is preferred to an alternative scarcity of a
greater range of commodities. At this point there will inevitably
arise questions concerning the rationality of individual choice,
questions which had been summarily suppressed by a cavalier
acceptance of the validity of assumptions ve and six set out in
the third section back. But these questions, whatever their im
portance, cannot be settled by deductive analysis.
The equality of price to marginal cost creates confusions in the
minds of many concerning the question as to whether or not costs
which are xed in the short run are not variable in the long run and
hence should be covered by price. They ask whether price should
not exceed short run marginal cost since the latter excludeselements
of long range variable costs. There are two confusions revealed by
this question. First, marginal cost is sometimes treated as a part
of total unit costs, which it is not. Marginal cost is the dierence
between costs in two situations and cannot be identied in general
with given components of cost, labor, materials, etc. Sympto
matic of this confusion is the statement that competition tends to
drive price down to prime or variable costs so that full costs are not
recovered. Of course, pure competition (which is not really im
In certain special ses consumer's surplus may be employed to describe nite
inequalities. But these cases are rare, and in any se we are rather better off if we use
direct methods.
plied in the prekus sentence) imohes price equal to marginal
cost, which may far exceed full unit costs dependmg Upon the scale
of output and the level of price Not only did Allred Marshall
speak loosely on this matter but so great a contribution as J M
Clarks Economics of Ozerhead Costs errs at least in expOSition in
its diagnOSis of the role of overhead costs m the breakdown of
competitite theory An examination of the character of agricul
ture, the last stronghold of pure competition, would show a high
level of overhead costs w1th no tendency to ard the breakdown of
competition Asxde from non imitable differentiation of product
the prinCipal cause of dismtegration of competition is decreasuig
long run unit costs of each firm at levels representing large fractions
of total demand for the product. this n ould be true even if all goods
could be produced to order
Marginal cost is not part of cost which must be met, and the
equality of price and marginal cost has nothmg to do With a re
covery of full costs determination of fair return on investments
correct imputation of factor shares. etc Its purpose is to secure
correct factor allocation. and at 01danomalous product allocation
This is accented if one conSIdersa wavy cost curve or a cost cune
showmg a corner at the pomt of capacrty (long run or short run)
At the latter pomt marginal cost is undened. or, if you prefer, it
is anywhere from a finite amount to innity There is no clue as
to proper price determinatlon, but thus is of no consequence so long
as productlon does actually take place at the capaCity pomt
Similarly as demand uctuates for a commodity u hose cost curve
shows waves because of Iumpmess, etc , the correct price charged
Willvary greatly, rather than let a train set out Witha Singleempty
seat. the price should fall to zero It is no argument against this
statement that such pricing may not recover full costs or that
passengers Willtend to wait until trains are least crowded What
could. be. more desirable. meJa} than the. evening, out of. trafc?
Does not a railroad company which gives out passes to its own
employees often follow the sound prmcrple that they may ride as
they please so long as they add nothing to costs? InCidentally. It
Will be seen from this discuSSion that where small changes in de
mand create great changes socnal cost, posted, administered.
relatively stable prices are not optimal devrces
The second and less important confusron involved in the state
ment that price should exceed short run marginal cost stems from
the fallacious belief that long run marginal cost inclusive of
changes in long run variable factors" is greater than short run
marginal cost. Actually, we know from the VVong-Viner-Harrod
envelope theorem that they are equal for instantaneous rates of
change if production is taking place at the anticipated level. For
an actual, nite, forward movement from this point, short run
marginal cost exclusive of the xed factors is of course greater
not smaller than marginal cost when all factors are varied opti
mally. Whether or not output is at the anticipated level, price
equal to short run marginal cost is necessary in order that the
existing plant be used optimally; the relationship between price
and long run marginal cost is relevant to a decision as to whether
to change the size of plant, when that decision arises. If one
dropped the rst order statement of optimum conditions and re
placed them by the general inequalities which must be satised,
it would be obvious that the optimum position must satisfy a
different maximum condition (inequality) for every alternative
which can be contemplated. Thus, it must not pay to make a
small step forward or backward, a middling step, a sizable step,
a huge step, to shut down completely or open a new line, etc., etc.
Each of these implies a relationship between price or revenue and
entirely different measures of marginal or differential cost.35
INTERPERSONAL OPTIMAL CONDITIONS
In the last two sections an attempt was made to derive as
general conditions as possible with a minimum of controversxal
assumptions. Nevertheless, we have seen that it is not posmbleto
deduce a unique equilibrium unless we have more to bu11dupon.
This is only as it should be, for intuition assures us that there can
not be an optimum position which is independent of the exact form
of the W function. Even if all of the necessary conditions of
production and exchange are satised, we are short as many equa
tions as there are individuals in excess of unity. In a Robinson
Crusoe world where there is but one indiv1dual (pace Friday !)there
Once this fact is mastered, the whole question of "allocation of perfectly joint
costs" is seen to be a false and irrelevant problem for either a rm or socnety. _It derives
its present day cmpiriml importance from the intrusion, not necessarily irrationally, of
full cost" considerations into price formation, and to problems ol government regula
tion, e.g., T.V.A., Tariff Commission, etc.
244 FOUNDATIONS OF ECOAOMIC ANALYSIS
are no mdlvxdualsexcess of umty and the ethbrlum 15umque
But as soon as we have more than one mdmdual our prenons
condmons s1mplyassure us that we are on the generahzed contract
locus from Wthh there are no possxble movements which are
advantageous to each and every Indrvndual
An Innity of such posrtzons eusts ranging from '1 Situation
whh all of the advmtage xsenjoyed by one mdrwdual through
some sort of compromise posntlon to one tn which another 1nd:
vzdual has all the advantage Wxthout 't well dened W function
1e thhout assumptlons concerning mterpcrsonal comparisons of
utlhty lt ts unpossnble to dec1de wlnch of these pomts :5 best In
terms of a gwen set of etlncal notlons wluch dene a Welfare
functzon the best pomt on the generalized contract locus can be
determined and only then
All of thls may be formulated mathenntncally as follows The
optimal condmons of production and change as gtxen tn equa
trons (26) whlch summanze the argument of the prcvnous two sec
trons permit us to reduce the low] of mdetcrrnmacy of the system
to an ImpllClt equation between the lcxels of hell being of the
different Inclwxdualsm the system Tina may be \\ ntten as
P(U1 U2 U) = 0 (27)
Thxs says that we may speedy at wrll the levels of ell bemg of
all but one of the mdwnduals and the last ones me]!being rs
umquely determmed The essential shape of thns Posszbzlztyfum:
non depends of course upon the technological and other assumed
restraints of the system as well as upon the tastes of the dlerent
mdn 1duals Notatlonally the same xmphcnt functlon can be
written m many dl'erent ways but the locus m questron ISxnvarl
ant under these purely terminological changes One should note
however that smce the exact numencal form of e1ch U ISarbltra'Y
no Importance attaches to the curvature propertzes of thls locus
If we have a gwen demte welfare function then we are to
mammxze
w = wwl U) (28)
subject to the above constralnt The rst order condition Of
ethbnum takes the form
W P
w: = E (: J =1 s) (29)
WELFARE ECONOrl-{ICS 245
These conditions, like the correct secondary conditions which need
not be written explicitly, are independent of purely notational
ambiguities involved in the welfare function, the possibility func
tion, and the selection of particular cardinal indices of individual
utility.
Equations (29) give us the missing (s 1) conditions of equi
librium, and our equilibrium is seen to be nally determinate. It
is not at all necessary to take the expositional path followedabove.
Thus, Professor Pigou, who does not hesitate from the very be
ginning to make inter-individual comparisons of utility, moves
directly via a maximization of (11) subject to (12) to the same equi
librium which Pareto-BaroneLemer reach after they have ad
mitted into their systems interpersonal comparisons of utility.
And as will be discussed at greater length below, they end up with
no optimum and no nal denite conclusions if they are unwilling
to permit others to introduce such ethical considerations.
The real signicance of their analysis lies in the fact that they
provide a relatively easy opening for the introduction of ethical
notions. For it follows from their analysis that all of the action
necessary to achieve a given ethical desz'deratummay take the form
of lump-sum taxes or bazmties. These may be given in the form of
abstract purchasing power or in kind, it being necessary, however,
that the latter should be freely exchangeable against other goods.
Thus, instead of having to decide how much to allocate of each and
every good to each individual, the ethical authority need only
decide on the allocation of nal incomes between individuals.
Working directly from a well-dened (ordinal) welfare function,
we require that W be maximized subject to the physical constraints
of the system. We may express the nal conditions in many
different forms, with or without Lagrangean multipliers. One such
compact method which avoids as far as possible duplicating the
exact form (but not the substance) of the conditions of the previous
section is given as follows:
-..._6__= <=L~un>
st x. GX,- (J = 1, - - -, m) (30)
6W _.'._____-._-._...
3W 6W 6W6X1 ___
6W 6X" .
Not only are these equatxons very snmllar to the famous Note
XIV whrch summanzes much of u hat xs best Marshall s Prm
czples, but they are cast almost exactly m the form followed by
Professor Plgou m hls Economzcsof Welfare In words, two condr
tlons are Implied rst, the margmal soczal utzlzty (dzsutzhty) of the
some good (servzce) must be equal for each mdwzdual, second, each
factor of production must be dwzded among dt'erent posszble uses so
that tts mdzrect. derwed margmal saczal utility must be the same m
every me and equal to mmargmol soczol dtsutzlzty
We may leave to the reader the task of showmg the necessary
modtcatrons 1f the supply of a factor IS Inelastic, If a factor of
productzon 15not mdlerent between dlfferent uses If there lS]omt
productlon, nf there are external technological economxes or dlS
economies (so that the production functton of a good contams m
xt factors of productxon devoted to other uses), etc By dmdmg
the above equattons through by any smgle partial denvatxve. they
may be thrown mto a form which ISIndependent of the part1cular
cardmal representatlon of W
Whlle the Lerner productlon condltlons are contained m the
above set of equatlons the exchange condltnonsare not However,
If we make the 1nd1v1duahst1cassumptlons ve and 81xabove so
that the welfare functxon takes on the spectal form gwen m equa
tron (13). then m Vlr-tueof the 1dent1t1es
6W W U'
xJ= _U'x: (31)
the exchange condmons of the prevnous section are also Included

m these equations Thus the production and exchange condttrons
whlch constxtute the new welfare economxcs" are Included m the
old, but are themselves Incomplete
If we substltute the last Identltles m the fundamental equatlons
(30), they can easzly be thrown mto the production and exchange
cond1t1onsg1ven m (26) plus the new Interpersonal element guaran
teemg us that the dlstrxbutlon of Incomes and lump sum allowances
IS optimal Thls takes the mathematncal form of equahty of
margmal soc1alutlhty of Income (expressed ln terms of any good)
for all mdwnduals
93 aw
6x3! : 633.2_
aw
xl.
)
HLLI'A KE L:CUNUMI CS 247
The statement that any individualistic ethical optimum can
be effected by lump-sum taxes is a theorem, not an axiom. For its
proof we require only that the individual indifferencecurves be such
that they can be swung into nal equilibrium by operating upon
their incomes, letting them trade mechanically at xed prices.
The vast importance of this theorem for social policy in a planned
or free enterprise economy is deserving of the greatest emphasis.
Yet its greatest signicance is in the realm of administration and
tactics. From the standpoint of the logic of welfare analysis it is
not fundamental. In fact it is not universally true.
First, the indifference curves of one or more individuals may
have curvature properties such that the individual who believes
or pretends to believe that prices are parameters beyond his control
will rush away from rather than toward the correct optimum posi
tion. The remarks in the section above dealing with production
which discuss the analogy between prices and Lagrangean multi
pliers and secondary conditions may be referred to here.
Second, and of more practical importance, the optimum position
may be reached by means of discriminating all or none offers,
sliding scales, etc., rather than by trading at uniform prices. The
important thing is the realization of appropriate marginal condi
tions rather than what happens to the intra-marginal units. (Com
pare on this point the discussion above of recovery of full cost as
a goal of pricing.) In a regime such as this, lump-sum allowances
may not sufce to realize the optimum point.
Third, it is not really easy to devise in practice a tax or subsidy
which is of a purely lump-sum character. A tax upon income
affects marginal decisions with respect to effort and risk-taking.
This is obvious. Less conspicuous is the fact that the usual ethical
notions compel us to set our allowances according to a mans
circumstances, which are partially the result of his own actions
and decisions. Analytically, the problem resembles that of deter
mining fair piece rates or of setting a fair handicap" for golfers
of different ability. We wish to equalize opportunity for all con
testants, but we do not want them to hold back from playing their
best because of a fear of losing their favorable handicap. Ideally,
the social managers would have to know the potentialities of every
individual; and to remedy the distortions of imperfect lump-sum
248 FOUNDATIONSOF ECONOIHC ANALISIS
taxes they would have to set a system of quotas and penalnes
based upon potentrahtues rather than upon performance
Thus, we mlght decrde that exeryone should have at least s
mxmmum :ncome, that Socrety nllmake Up the decnency betweer
what the less fortunate can earn and thls mmrmum Once thus 15.
reahzed by those who {all below the mmrmum, there rs no longer
an mcentne for them to work at the margm at least 1npecumar)
matenal terms Thns rs clearly bad socral pollcy, not because I
haxe a vulgar preJudxce ln favor of work and agamst lelsure On
the contrary, the Increases m real mcome 1n the years ahead prob
ably wrll be spent m consrderable degree on lcrsure It :5 wrong
because xt forces the rest of soc1ety to glve up IEISUIe Its fault hes
m the fact that the 1ndw1duals allowance ISnot &lump sum one
l t vanes rm ersely wrth hrs eort, and thus penahzes effort lt
would be tolerable only ll a small percentage of the p0pulatlon
were below the minimum or If we could rely upon new motwatrons
ln the Utopra of the future I, myself, put consrderable stock
the possnbrhty of changmg conventlonal patterns of motlvatron
However, thrs wzllnot prowde comfort to those n ho Wishto utzhzc
a parametnc prncmg system wrth algebraic lump sum allowances
smce these same consnderatlons undermine the 1nd1v1duahstrc
assumptrons upon whrch thexr analyses rs based
Fourth because decnsrons need only be made concerning m
comes, the problem of formulatmg polrtrcal slogans and behels
whrch \ull command wrde approbatnon rs 51mphed However de
srrable thrs may be from a pohtlcal pomt ol vrew, It must never be
forgotten that from a consxstent ethrcal pomt of v1ew decrsxons
should be made concernmg the welfare lunctnon 1tself Behefs con
cermng'the drstnbutron of Incomeare dem/am e rather than funda
mental Except unthe adrmttedly unreahstlc case where all tastes
are rdentnl, settmg up such belxefs as goals xs equwalent to ac
ceptmg a "shxbboleth" and to embracmg an ambuguous unden
" The clash between equrty and meentwe :5 not unimportant for the present stage
of modern ptmlum To an Important rst approxrmauon the adverse eectso! h gb
margmal rate- of taxation stem not from the level of the rates provaded they are less
than 100 per cent but from the curvature of the tax formula whereby game mcrmse
tax lxabrlny more than equivalent losses reduce tax habrluy But the essence of pro
grtsston m taxatuon and m moorne redrstnbutnon :s curvature The only solution
les m poohng rrsks and m drafung tax leguslauou so as to produce a mammum ol
veragmg oser tune If this does not sufce equltv may stnll be worth consademblc
\avmdable costs
able welfare function. Among other things such a procedure in
volves making a shibboleth of the existing distribution of relative
technological scarcities of goods. (Of course, whether this is a bad
practice is not in itself part of the content of welfare economics,
which does not presume to deduce appropriate beliefs. However,
welfare economics can legitimately point out the implications of
different ethical propositions.)
NEW VERSUS OLD WELFARE ECONOMICS
While in a real sense there is only one all-inclusive welfare eco
nomics, which reaches its most complete formulation in the writings
of Bergson, it is possible to distinguish between the New Welfare
Economics, which involves roughly the contents of the sections on
production and exchange and which makes no assumptions con
cerning interpersonal comparability of utility, and the Old Welfare
Economics which starts out with such assumptions. Roughly it
is the distinction between Pareto and Pigou. From our above dis
cussion it will be clear that the former is included in the latter, but
not vice versa.
Strictly speaking there is no opposition between the two points
of view. However, it is not uncommon for expositors of the new
set of doctrines to imagine that their results have signicanceeven
if one is unwilling to make any ethical assumptions. In fact this
belief is almost a necessity for anyone who has taken seriously
Robbins' dicta concerning the inadmissibility of welfare economics
in the body of economic theory. Once it is apprehended that the
latter notion is a delusion, the need to believe in the signicance of
the production and exchange conditions divorced from the inter
personal conditions will disappear. Nevertheless, it may be de
sirable to appraise their signicance taken by themselves.
I pass over as being obvious from our discussion of the rst six
assumptions about the W function that it is not literally true that
the new welfare economics is devoid of any ethical assumptions.
Admittedly, however, its assumptions are more general and less
controversial, and it is for this reason that it gives incomplete
necessary conditions, whose full signicance emerges only after one
has made interpersonal assumptions. To refuse to take the last
step renders the rst two steps nugatory; like pouring out a glass
of water and then refusing to drink. To say that the rst two
conditions should be realxzed, but that the thnd IS meamngless, 15
lnkesaymg that it does not matter whether or not a man has han-,
so long as xt :s curly'
A l1m1tedsugmcance remams for the new welfare economlcs1f
we hold that a welfare functlon 1s denable but undened. and If
we look for condltlons which hold true umformly for all p0551b1e
demtnons It cannot tell us wh1chof any two sntuatlons lSbetter,
but 1t can occasxonally rule out one gwen sntuanon as bemg worse
than another m the sense that everyone ls worse off It cannot tell
us when society really has a chonce between two gwen s1tuat10ns
Most Important of all 1t cannot tell us that a movement about
Wthll 1t can gtve a determmate answer 15better than a movement
about whtch 1tcannot gwe an answer If we are at a pomt not on
the P0551b1htyFunctlon, 1t assures us that there exnsts a stxllbetter
pomt But It cannot assert that a gwen pomt on the Posmblhty
Functlon IS better than all. or many. pomts not on the P0551
blllty locus
Concretely, the new welfare economlcs xs supposed to be able
to throw llght on such questtons as to whether the Corn Laws
should have been repealed The act would help many 1nd1v1duals,
but hurt landowners It 15usually assumed, although not neces
sarrly quzte ngorousl; , that free trade can be expected to lead to a
new sntuatlon 1n Wthh those who are beneted can afford to
brnbe those who are hurt It IS thought that thts constxtutes a
case for free trade Actually. thzs leads to no posmve gurde to
act1on For the new welfare economics cannot state demtely
The Corn Laws should be repealed, and the landowners should be
compensated " Occasronally such a statement 15 1mp11c1tm a
writer's formulation, but nt 15clear that th1s mvolves an unwar
ranted assumptton as to the adequacy of the status quo, an assump
tlon winch the new welfare economics holds to be neither nght nor
wrong, but meamngless
On the other hand, the new welfare economics cannot make the
denite statement The Corn Laws should be repealed, and the
owners should not be compensated " It can only assert m a nega
twe way Twere better that the Com Laws be repealed, and
compensatlon be pa1d, 1f necessary " It gues no real gunde to
actlon Nor does our experience w1th man as a socnalammal sug
tthat one can safely predtct, as a factual matter, that educated
and intelligent men of goodwill" in point of fact tend to move to
the generalized contract locus. As an empirical statement of fact
we cannot agree with the assertion of Edgeworth that bilateral
monopolists must end up somewhere on the contract curve. They
may end up elsewhere, because one or both is unwilling to discuss
the possibility of making a mutually favorable movement for fear
that the discussion may imperil the existing tolerable status qua.
Scitovsky 37has attempted to remedy certain decienciesin the
usual statement of the new welfare economists by developing a
double test of the desirability of two situations, in which the dis
tribution of income of the new situation is treated symmetrically
with that of the old situation. While his is an improvement on
certain earlier statements, it is so primarily in a negative way, to
the extent that it delimits the scope of the new welfare economics.
His own positive formulation is also a twere better" statement,
and not a positive guide to action. Such statements are not with
out value, but they are not at all substitutes for the policydictates
which stemmed from the old welfare economics.
Within the limited scope of the 'twere better" statements
there is an ambiguity which Scitovsky alone seems to have sensed.
The term situation can mean a number of different things. It
can refer to an actual position reached by every individual prior to
the repeal of a tariff, and to the new situation which would actually
be attained by each individual after new action is taken, i.e., repeal
with varying or no amounts of compensation. Or it can mean the
totality of all possible positions available with the Corn Laws un
repealed, and the totality of all possible positions with the Corn
Laws repealed. The latter is the more signicant sense, and can
be given signicance in terms of the Possibility Function developed
above. Thus, a perfectly legitimate twere better statement
can be made as follows: Technological change can make everybody
better off in the sense that it will shift the Possibility Function out
wards." 33 We cannot deduce from this the dictum: Techno
logical change is a good thing, since the introduction of techno
logical change will in fact mean a vector movement from the old
"' T. Scitovsky, A Note on Welfare Propositions in Economics," and A Recon
sideration of the Theory of Tariffs," Reviewof Economic Studies. IX (1941-42), 77-88
and 89110. . __ _ _ ,
" In effect Scitovsky recognizes the case where the Possnblhty Function is given a
twist rather than everywhere being shifted outward.
252 IOUNDATIONS OP ECONOMIC ANALISIS
posmons of every 1nd1vrdual to their new posmons whxch can
hardly bode good to all
At tlus pomt the new welfare economlst wnll fall back on the
vague statement that probably technologtcal change wrll result
m good on balance " Not only 15tl'us argument on Jts probabxhty
s1debased upon 1gnorancerather than knowledge, but xtISmeamng
less as well unless the admlss1bxhty of a W functlon ISadnutted
In closmg I should hke to pomt out that the pomt of vxewWthh
I have been calhng the new welfare economics 15 really only a
cancature On the whole Pareto, Barone. Hotelhng, and Lerner
sxdestepped mterpersonal Judgments rather than demed them
Bergson synthesmes the venous aspects, as does Lange Whether
Kaldor and chks are open to the cntlcnsm along the hnes hxch
Strgler has recently advanced, the reader must decxde
CONCLUSION
In the present chapter I have attempted to gwe a bnef but
fan-1ycomplete survey of the whole eld of welfare economxcs It
would be possrble to develop the subject further m many respects
and to consrder a number of related problems HoweVer, hmnta
trons of space preclude thus, and I shall content myself wzth two
nal remarks
Fxrst, what 15the best procedure If for some reason a number
of the opt1mum condxtlons are not reahzed? What shall we do
about the rem-unmg ones Wthh are 1nour power? Shall we argue
that two wrongs do not make a rlght and attempt to satxsfy
those we canP Or IS xt possnble that {allure of a number of the
conditions necessuates modxfymg the rest? Clearly the latter
altematnve ls the correct one A gwen divergence a subset of
the optimum condltrons nece551tatesalteratlons m the remammg
ones Thus, m a world where almost all mdustnes are producmg
at marginal soc1al cost less than _once _(cltherbecause of monopoly
or external economies) xt would not be desrrable for the rest to
"N Kaldor Welfare Propos nous Economl Ecanomu: Journal, XLIX
(1939) 5495552 } R Hrclcs Foundattons of Welfare Econormcs Eco/tam: Journal
XLIX (1939) 696712G j Sugler The New Welfare Economrcs Amman: Eco
nom:: Rmew XXXIII (19-13) 355359 P A Samuelson Further Commentary on
Welfare Econom Cs Ame/acct: Econamzc Raman; XXX!!! (1943) 604-607 See also
P A Samuelson Welfare Economn and International Trade Amencan Economie
XXVIII (1938) 261-266
produce up to the point where marginal cost equals price. Neither
would it be quite correct to seek the same percentage or absolute
divergence from the optimum conditions in each case; although in
this particular example, if the elasticity of the suppliesof the factors
of production were zero, the proportionality of prices to marginal
cost would be as good as the exact equality. Still another example
to show that failure of some conditions necessitates alteration of
the rest is provided by the possibility of increasing welfare by de~
liberately selling below marginal costs to groups with a high mar
ginal (social) utility of income. Given a faulty distribution of
income, this can improve the situation, although it would be still
better to have the full optimal conditions realized.
The lastpoint consists of the warning that the introduction of
dynamic conditions into our analysis necessitates a considerable
change in the statement of optimal conditions.40 The difference is
not one of principle, but it is nevertheless important. Judged
purely on statical grounds, monopolies or a patent system may
appear as unmitigated evils, and certainly inferior to atomistic
competition and free trade. But in a dynamical world these judg
ments may have to be reversed; viz. the infant industry argument
for protection, the stimulus to large scale research which only a
monopolist can afford, the (alleged) necessity to hold out incentives
to inventors, etc. Indeed the measure of support which capitalism
commands is most importantly related to precisely these factors
of development. .
In admitting the. superiority of monopoly to atomistic competi
tion in certain respects, we do not imply that it is the best possible
organization of an industry. There necessarily exists a still better
third alternative, which may or may not be less Utopian than the
restoration and maintenance of atomistic competition.
Arguments which allege that individuals, under mpitalism or socialism, decide
what goods they shall consume, but are not capable of making a correct decision wrth
respect to saving seem to suggest that there is a qualitative difference introduced by
dynamics. I refer to the argument that wealthy men are necessary in order 10 PfOVlde
for saving and capital formation, and that in a socialist state the government should
decide what the appropriate amounts of ptal formation should be. If one speaes
a W function satisfying the rst six assumptions and includes in it as separate-variableS
future goods and services, both of these views are wrong whether there is pxhsm or
socialism. However, especially where problems of the present and future are concerned,
modern philosophers balk at the assumption that what people think is best for them
really is. Naturally, when the individualistic assumption six is dropped, our conditions
are altered. But the same would be true if we denied this assumption in a statical world.
CHAPTER IX
THE STABILITY OF EQUILIBRIUM: COMPARATIVE

STATICS AND DYNAMICS
INTRODUCTION
IT WASANachievement of the rst magnitude for the older mathe

matical economists to have shown that the number of independent
and consistent economic relations was in a wide variety of cases
sufcient to determine the equilibrium values of unknown economic
prices and quantities. Since their life spans were only of nite
duration, it was natural that they should have stopped short at
this stage of counting equations and unknowns. It remains to be
explained, however, why in the rst quarter of the twentieth cen
tury economists should have been content with what was after all
only preliminary spade work containing in itself (at least ex
plicitly) few meaningful theorems of observational signicancesuch
as could even ideally be empirically refuted under any conceivable
circumstances.
It is the task of comparative statics to show the determination
of the equilibrium values of given variables (unknowns) under pos
tulated conditions (functional relationships) with various data
(parameters) being specied. Thus, in the simplest case .of a
partial-equilibrium market for a single commodity, the two inde
pendent relations of supply and demand, each drawn up with other
prices and institutional data being taken as given, determine :DY
their intersection the equilibrium quantities of the unknown price
and quantity sold. If no more than this could be said, the econo
mist would be truly vulnerable to the gibe that he is only a parrot
taught to say supply and demand. Simply to know that.there
are efcacious laws determining equilibrium tells us nothing Of
the character of these laws. In order for the analysis.to be useful
it must provide information concerning the way in which our equ1
librium quantities will change as a result of changes the param
eters taken as independent data.
257
258 FOUNDATIONb 01 ECONOMIC ANALlSIS
In the above xllustratxon let us consxder tastes as a changmg
parameter mfluencmg only the demand curve W111an Increase 1n
demand raxseor lower pnce? Clearly the statement that before
and after the assumed change prxce xs determined by the mter
sectxonof supply and demand nges us no answer to the problem
Nothing can be said concernng the movement of the Intersection
pomt of any two plane curves as one of them slnfts And yet most
economists would argue that xn a vude varlety of Circumstances
thxs questxon can be gwen a denite answernamely that pnce
wxll Increase
How xsthus conclusron denved? Por few commodxtxeshave we
detailed quantitatue empmcal Informatron concerning the exact
forms of the supply and demand cunes even xn the nelghborhood
of the equxlxbrxumpoxnt Not only \xould large amounts of txme
and money be necessary to get such Inlormatmn but m many cases
xt xs practically xmpossxbleto deme useful empmcal Information
concermng what would happen xf vxrtual changes xn pnce con
fronted the demanders or the supphers
Thxs xs a typical problem confronting the economtst xn the
absence of precnsequantrtatne data he must mfer analytlcally the
quahtatwe direction of movement of a comple\ system What
httle success he has hltherto achrex ed can be cI'xSSxed large part
under two headings (1) theorems proceeding from the assumptnon
of maxlmxzmg behauor on the part of rms or mdnxduals and
(2) stabnhty conditions relatmg to the mteractron between economxc
units Although Inadequately explored until comparatwely re
cently the rst type of condxtxonsxsbest Lnonn and I not be
dealt mth here except mcrdentally As wzllbecome ewdent later
hon exer from certam pomts of \ xe they can be tted m as specral
cases of the second set It xs the central task of thxs chapter to
show how the problem of stability of cquxlxbrxumxsmtxmately txed
up thh the problem of derxvmg fruxtful theorems m comparatne
statxcs Thus duahty constntutes what I have called the carre
spondence prmczple
COMPARATIVESrArxcs
The problem may be approached m full generahty by consxder
mg 7:unknown vanables (x1 x) whose equxlxbrxum\alues are
\ be determlned for preassrgned values of a parameter a We
STABILITY AND COMPARATIVE STATICS 259
assume n independent and consistent, continuously differentiable
1mphc1t relations involving some or all of the unknowns and the
parameter a; or
f(x1,---,x,.,a)=0. (13=1,--,n) (1)
These determine a set of equilibrium values
x," = g,(a). (2)
We wish to determine the sign of
%fj
= ma). (s)
Differentiating (1) totally with respect to a, we can express this as
dx.
da=_ :
T j: AH ' @
where the subscripts indicate partial differentiation,
fx f2: ... fn
A=fff =Ife'l.
f 1" f 2" f n"
and Ai;is the cofactor of the element in the jth row and ith column
of A.
Unless some a prion" restrictions are placed upon the nature of
the elements involved in these determinants, no useful theorems
can be derived. Each unknown derivative depends upon an
n(n + 1) innity of possible values. If the various determinants
were expanded out, a sum of n! terms would appear in the denomi
nator and in the numerator. Regarded simply as chance drawings
taken at random from a hat, the probability that the signsof these
would all agree would go rapidly to zero as the number of variables
increased. Fortunately, as will be shown, the analysis of the
stability of equilibrium will aid in evaluating these complicated
expressions.
If for a given value of a = a1, there exists a solution (xr, -- -,x..'). and if the
matrix [af/ex,] is ol' rank n in a neighborhood of (x), then by the implicit-function
theorem equations (2) represent single-valued continuously differentiable functions in
a sufciently small neighborhood of (ax. x).
In the Simpleexample of supply and demand alluded to above
our unknowns are (p. q), and our equnhbnum system can be wntten
gD(pa)=0 (D.>00,<0)
qS(p)=0 (5)
where a ts a parameter of shtft representing taste and D,, 1%
usually assumed to be less than zero Also
31_ . __1__
$ Da Spo __ Dpo v (6)
(19:
= D o __.P_. (7)
Whether or not pnce wdl Increase when dunand Increases IS

seen to depend upon the algebralc dncrence bet een the SIOpes(re
ferred to the puce ans) of the demand and supply curves at the
cqunllbnum pomt Quantity wdl mercase only If the slope of the
supply curve 15of the same 51gnas thus algebralc dszerence If the
system 15stable m the sense of Walras xt can be shown that the
supply curve must have a greater algebralc slope than the demand
curve so that puce wnllnecessanly Increase. the change m quantnty
xsnecessarily of amblguous Slgndependlng upon whether the supply
curve 15posxtwely melmed on so called backward rlsmg "
STABILITY AND DYNAMICS
Before denvmg exphmtly the Walrasran stability condmons re

ferred to abo, I turn to a dtscussnonof the meaning of a stable
equthbrxum Tlus Ill be found to presuppose a theory of dy
namxcs namely a theory which determmes the behavxor through
tlme of all vanables from arbltrary 1n1t1alconditions If we have
gwen n v1nables [an(t), , x(t)] and ?: functnoml equatnons of
The dzsuncuon suggested bv 'Slrkahn between forward falling and back ard
115mg negauvely 1nclned sapply curves while suggestwe does not rest upon a dynam c
analysrs of the attamment of equll br um and so does not adequately come to gr ps thh
the problem m all 1tscomplexnty Cf R F Kahn The Elasumy of Substitution and
the Relative Share of a Factor Remap of Economic Studres I (1933) 72 73 315 !
haldor A Classxlactory Note On the Determmateness of Equ lbnum Rene-wof
Economtc Studies I (1933) 122 136
The suggest Onof Professor Vmer that the latter type of curve gives the maximum
amount Iorthcom ng at a gwen pnce while the former does not Willreceive amp catnon
In the course of this argument
the general form
t t ,
F*[x1('r). 372(1), . . ., x,.(f), t] = O, (5 = 1, ___, n) (8)
then their behavior is determined once certain initial conditionsare

specied.3 Examples of functional equation systems are given by
sets of differential, difference, mixed differential-difference, integral,
integro-dilferential, and still more general systems. Following the
excellentterminology of Professor Frisch,4 stationary or equilibrium
values of the variables are given by the set of constants (x10,. . ., xno)
which satisfy these equations identically, or
t l I
Ff[x1,.2, xn,t] = 0.5 (i = 1, --,n) (9)
If the system has always been in equilibrium up until time t, it

will subsequently continue in equilibrium. However, the equi
librium values (xf, - - -, x")may be attained or even be maintained
for a nite period of time, and yet because of generalizeddynamical
inertia it need not (and in general will not) remain in equilibrium
subsequently, but may well overshoot" the mark.
The equilibrium position possesses perfect stability of the rst
kind if from any initial conditions all the variables approach their
equilibrium values in the limit as time becomes innite; i.e., if
lim nc.-(t) = x., (10)
106:
regardless of the initial conditions. Alternatively, it is sometimes

stated that an equilibrium is stable if a displacement from equx
3What constitute initial conditions depends upon the nature of the functional equa
tions. For dierential systems only values of the coordinates, velocities, and higher
derivatives at some initial time need be specied. For differenceequations dened only
for integral values of t the same is true, where differences replace derivatives- I the
Keneralcase values of the variables over a continUOustime interval, 995.5%stretching
back to - oo,are required to constitute a complete set of initial conditions. .
4R. Frisch, On the Notion of Equilibrium and Disequilibrium," Reviewof Econonm:
Studies, III (1936). 100105.
Of course, such a set need not exist. Thus. the simple system
dx
-= et - x
dt
has no stationary equilibrium values since ez - x = 0 has no real roots. Similarly.
dWt = l denes no stationary equilibrium position.
librium is followed by a return to equilibrium A displacement is
equivalent to an arbitrary change in the initial conditions and is
p055ibleonly if some of our functional equations are momentarily
relaxed or if our system is enlarged to include impressed forces
or shocks
Stability of the rst kind m the small eusts if for suFCiently
small displacements the eqmlibriuin is stable Stability in the
small is contained Within perfect stability but not Vice versa A
system may be stable for small nite displacements but not for
large displacements Nevertheless, stability in the small is a
necessary condition for perfect stability and Willbe analyzed here
in greatest detail
It should be pomted out that no conservative dynamical system
of the type met m theoretical physws possessesstability of the rst
kind If one displaces a frictionless pendulum, it Will osmllate
endlessly around the pOSitionof stable equilibrium Its motion
is bounded, however and it never remains on one Side of the equi
librium pOSition for more than a nite time interval Such be
haVior may be characterized as stabzlzty of the second kmd or as
stability in the second sense As before a distinction can be made
between stability of the second kind in the small and complete
stability of the second kind For the most part in the present
investigation I shall be concerned mth the problem of stability of
the rst kind
The equations of comparatiie statics are then a speeial case of
the general dynamic analySis They can indeed be discussed ab
stracting completely from dynamical analySis In the history of
mechanics, the theory of statics was developed before the dy
namical problem as even formulated But the problem of sta
bility of eqmlibriuni cannot be discussed except With reference to
dynamical conSiderations, however impliCit and rudimentary '
'We find ourselves con'ironted with this paradox in order lor the
comparative statics 2111813518
to yield fruitful results, e must rst
A dynamical system into which friction is introduced Via a diSSipation function
may enjoy stability of the rst kind On these and kindred matters see G D Blrkboff
Dymnmcal System: (New York 1927)
7This is seen to be involved in the Virtual work analySis and in the minimum
potential energy condition characteristic of a stable station! ( stationary ) equilibrium
non
STABILITY AND COA/[PARATIVE STA TICS 263
deve10p a theory of dynamics.8 This is completely aside from the
other uses of dynamic analysis as in the studies of uctuations,
trends, etc. I turn now to some illustrations of these propositions.
I. In the literary explanations of the process by which supply
and demand are equated, the assumption is usually made that if
at any price demand exceeds supply, price will rise; if supply ex
ceeds demand, price will fall. Let us state this more precisely as
follows:
. d
p==wrwa=MEm-S@l un
where H (0) = O,and H > 0.
In the neighborhood of the equilibrium point this can be ex
panded in the form
15: )(Dpo_ Six? _ Po) + ' ': (12)
where A = (H ) > 0, and where terms involving higher powers of
(p p") are omitted. The solution of this simple differential
equation for initial price ['3at time zero can be written at sight:
pw=w+@wwwwm no
If the equilibrium is to be stable,
lim PU) = P- (14)

This is possible if, and only if,
D,, 5 O. (15)
If in what follows we rule out neutral equilibrium in the large and
in the small, the equality sign may be omitted so that
D; 5; < 0. (16)
If the supply curve is positively inclined, this will be realized. If
it is negatively inclined, it must be less steep (referred to the price
axis) than the demand curve. If our stability conditions are
realized, the problem originally proposed is answered. Price must
rise when demand increases.
' The point made here is not to be confused with the commonplace criticism of
comparative statics that it does not do what it is not aimed to do, namely describe
the transition paths between equilibria.
264 I'OUNDATIONS OF ECONOMIC ANALlSIS
ll These so called Walnsran stabxht} conditions are not neon
sanly the only ones ll alternative rlymmlc models an. postu
lated completely dlffcrcnt conditions are deduced which m tum
lead to alternatrve theorems comparatne statues
Thus m Marshall 5 long run theory ol normal price the quan
tlty supphed ISassumed to adjust Itself comparatwely slow]; If
demand pnce exceeds supply pnce the quantzty supphed wxll
Increase Preservmg our notation of equations (5), remembenng
that quantity rather than pnce IS regarded as the Independent
\artable and neglectmg lngher order terms ue have the followmg
(hfferentl ll tqu mon
1 l
q=k(u,o-s,~)(9-) (k>0) (17)
whose solution 15
90) = q" + (q q)e*' "5' (18)
If the equlhbnum ISto be stable
1 l l S D,,
H,?S?=7' <0 (1)
: e the demand cune slope referred to the quanttty ans xsless
algebraically than that of the supply cune Smce the demand
cune ISnegatn ely 1nchned
S,
> 0 (20)
Relemng back to equatlons (7) ne see that \Iarslnllnn stabxhty

condxtlons require that quantity Increases hen demand lCfLaSLS
m exery case wlnle the change m pnce IS necessanly ambtguous
dependmg upon the algcbmlc srgn of the supply cune s slope
It rs to be pomted out that thus {omard fallmg supply cunc IS
not a true supply cune m the sense ol the amount forthcommgat
each hypothetncal pnce, although nt IS a true supply cunc m the
sense of bemg the locus of pnce quanttty pomts traced out by
fluctuations m :uicrcny steep demand cunes As such 1t xs !
reersnble long run relation
' A h stor cal error ISI! olvcd m the Identrfmt on of the above stab l ty oond t on:
v. th alfa: alleged contrast to those of Marshall Inch are shortly to be d scussed
Actually as far back as m the Pure Theory of Fare ; Trade Marshall dened stable
equ l br um m wluch a socalledbackward r s ng supply curve mas nvolved much I he
Walma an case
III. Still another dynamic model may be considered. It has
been held that for some commodities supply reacts to price only
after a given time lag, while price adjusts itself almost instantane
ously. This leads to the familiar cobweb phenomenon. Using the
same notation, our dynamic model takes the form of the following
difference equations,
Q: = SGH1).
Q: = DU) a). (21)
In the neighborhood of equilibrium
S o
(g. qe) = ( Dj,) (g q.) (22>
with the solution
5 a t
q.=g+<qq>(D:.)- (23)
Stability requires that
g
P < 1. (24)
If the supply curve is positively inclined, it must be steeper abso
lutely than is the demand curve, reference being made to the
quantity axis. In this case the approach to equilibrium is of a
damped oscillatory nature, every other observation being on one
side of the equilibrium value.
If the supply curve is negatively inclined, it must be steeper
referred to the quantity axis than is the demand curve, precisely
as in the case of Walrasian stability. The approach to equilibrium
is asymptotic. As in the Walrasian case, we can deduce the
theorem in comparative statics that price will necessarily increase
even though the change in quantity is indenite.
It is to be noted that a rst-order differenceequation is richer in
solution that the corresponding rst-order differential equation.
Not only does it admit of oscillatory solutions, but the stability
conditions relate to the absolute value of the root of an equation,
implying two distinct inequalities. Rememberingthat D; is nega
tive, the inequality of (24) can be written
D; < 5; < D,.
The new inequality tells us that any increase in output as a result
of an increase in demand cannot be so large as the increase output
resulting from an equivalent" increase in supply.
IV. Still a fourth dynamical model that has been considered is
that of Marshall in the Pure Theoryof Foreign Trade. Let Figurc2
represent the familiar o'er curves of two trading bodies (suppliers
and demanders) respectively. Equilibrium is attained at the in.
tersection (not necessarily unique) of two such curves If equi
lb !.
FIGURE 2 FIGURE 3
librium is displaced, country I is to act in such a way as to change

the amount of x. in the horizontal direction of its offer curve
(as indicated by the pointed horizontal arrows). Similarly coun
try 11 adjusts x, vertically in the direction of its oer curve.
Mathematically,
21 = HIEGCV _ 1'1].
(25)
I: =
H,[F(x1) _ x3],
Where II" > 0' IIt(0) = 00 and GO) _ x; = 0. F(x1) .. x, = 0
represent the statical o'ercurves of countries I and II respectively.
For H { = H 2and proper units, the following system of di'erential
equations will hold in the neighborhood of equilibrium,
351= _ (xl _ xl") + (G')(xz _ If).
(26)
133= (F')(x1 _ x1) _ (x2 " x2).
The solution takes the form:
x10) = 351+ knew + 16sz1". (27)
3:20) = 362+ knew + knew.
the k's depend upon the initial values (:il, 52)and the A'sare
roots of the characteristic equation
1x 0o
Cleary
1 DO) : (F') -(1)x = (28)
x = 1 :|: J('G')o'(F')'o._ (29)
The equilibrium will be stable if the real part of X is necessarily
negative, or
RO) < 0. (30)
If both (G') and (F ) are of opposite sign (e.g., if one has an elastic
demand, the other an inelastic demand), this condition will neces
sarily be satised. The solution will be oscillatory, but damped,
approaching equilibrium in a spiral as shown above in Figure 3,
and obeying an equation of the form
x.- = x. + e"(a.sin 6t + b,-cos 0t). (12= 1, 2) (31)
If both are positive, however (each with elastic demands), then
4(0)(F) < 1. (32)
(G)(F) < 1. (33)
and
(G') < (7%) (34)

In terms of the SIOpesof both offer curves referred to the xl axis
@
(dxl) 513)II.
> (de (35)
The equilibrium is approached asymptotically. .
If both curves are negatively inclined, stability requires
(dx1)1<
1322 dx] II .
(!) (36 )
Clearly, the general condition when the curves are of like sign can
be written
Ea ! a (37)
dxl I dxl II
and the approach to equilibrium will in every case be asymptotic.lo

"Somewhat paradoxically, in this se positious of stable equilibrium need not be
separated by positions of unstable equilibrium because of the possibility of complex roots.
The stability conditions derived here will be found, if translated
into terms of supply and demand curves rather than offer curves,
to impy differing and inconsistent conditions from those of the
preceding cases.
V. In the {our cases considered I have been concerned with
problems of stability of the rst kind Following a suggestion of
Dr. Francis Dresch of the University of California, let us suppose
that price falls not when the instantaneous supply exceedsdemand,
but only when accumulated stocks exceed some normal value, Q,0r
p=7\(Q"-Q)+ =xooxfo'sqadt. o>o> (38)

since the stock equals the accumulated di'erence between the
amount produced and the amount consumed. Differentiating with
respect to t, neglecting terms of higher power, and redening our
time units so as to suppress the dimensional constant A we would
have 5 = (D.: smtp po). (39)
whose solution is
P) : po + cle(JDO-5p-)3
+ Cze(mjl. (40)
the cs depending upon the initial price and price change.
Only if
D,, - S, < 0, (41)
can explosive behavior be avoided. If the above inequality is
realized, however, the square root will be a pure imaginary number
so that the solution takes the form of an undamped harmonic:
p) = b; cos \lS,
D,: + b: sin 45; -D,, + P. (42)
Thus, if we require our second-order differential equation to have
at least stability of the second kind, we come to the same theorems
in comparative statics as in case I.
There is at least one serious objection to assuming a nondamped
system of this kind. If there are superimposed upon our system
random shocks or errors, these will tend to accumulate so that the
expected amplitude of the cycles will increase with time. This is
well illustrated by the familiar Brownian movement of large mole
cules under the impact of random collisions. The molecule takes
a random walk," and its mean variance increases with the observa
tion time.u Before adopting a similar hypothesis in economic
analysis, some statistical evidence of its possible validity should
be adduced.12
We have now surveyed ve different dynamic setups and re
lated stability conditions, all referring to a simple one-commodity
market. Except possibly for cases IV and V, all are mathemati
cally trivial. Unaided intuition or simple geometrical methods
serve to reveal sufcient conditions for stability. They are of sig
nicance, however, because each has played an important part in
the history of economic science; and precisely because of their sim
plicity they provide a useful illustration of the general principle
involved. In the following sections I shall be concerned with more
complex problems.
THE STABILITY or MULTIPLE MARKETS
While it might be more elegant at this stage to develop formally
for general systems the fundamental principles illustrated thus far,
our foregoing discussion provides a very convenient opening for
an examination of a problem which has received considerable at
tention lately at the hands of Professor Hicks. In Value and
" Random shocks are not necessarily to be regarded as a nuisance. In their absence
friction might imprison the system at some xed level other than the true" equilibrium
level (friction being disregarded). Often random shocks serve to insure the realization
of average values nearly equal to the equilibrium ones, just as iron lings placed upon
a piece of paper over a magnet assume the lines of force of the magnetic eld when
gently tapped.
One n avoid an undamped system by assuming that price tends to fall not
only when stocks are large, but also when current supply exceeds current demand;
i.e., when stocks tend to accumulate. Then we have
I
p = a [0" -j; (qs qo)d1](q.s an)

or
; = (D; _ 511)?+ MD, " 5,911 (a, B > 0)
The equilibrium is stable only if
ROI) < 0,
or if
aw; - am -a (D; $;) = 0.
or if
D; 8; < 0.
This agrees with the condition of se I and the one just derived. In fact, each of these
is a special se when one of the coefcients vanishes. I:orintermediate values, the
solutions range connuously between damped harmonic motion and exPOnenUl
approach to equilibrium.
Capital, Chapter VIII and Mathematical Appendix, 21,he has at
tempted to generalize to any number of markets the stability con
ditions of a single market The method of approach is postula.
tional; stability conditions are not deduced from a dynamic model
except impliCitly.13 Propositions which are deduced here as
theorems are assumed as denitions of stability.
For a single market, according to Professor Hicks, equilibrium
is stable if an increase in demand raises price (This rules out in
the beginning cases II and IV.) For multiple markets equilibrium
is zmperfectlystable if an increase in demand for a single good raises
its price after all other prices have adjusted themselves; the equi
librium is perfectly stable if an increased demand for a good raises
its price even when any subset of other prices is arbitrarily held
constant (by means of a relaxation of other equilibrium conditions).
To test the neceSSityor sufciency of these criteria in terms of
a more fundamental denition of stability of equilibrium let us
make a natural generalization of the \Valrasian conditions of the
followingform: the price of any good ill fall if its supply exceedsits
demand, these each bemg regarded as functions of all other prices.
Mathematically,
? = _ H.(q.s' "' QD)
: 'IIIEqS1(pIo ! p) _ QD,(PI' 'pn)]
= H. ! 0810(1))_ P)) + ' (43)

where
0 98%? iPn) "' %(Pn - P) = " QG) ' " p) (44)
represent statical equations of supply and demand, a,; represents
the partial derivative of q. with respect to the jth price evaluated
at the equilibrium set of prices In general, a., # ao It S
It IStrue that on page 70 a hint of a dynamical process creeps into the discusSion
The approach to equilibrium seems to be regarded as taking place in nite steps at dis
crete intervals of time. i e . in accordance With certain difference equations Correctl)
stated. this argument would not lead to essentially di'erent stability conditions from
my 8)Stem of differential equations discussed below, as the later general discussion
V.ill disclose
If the demand and supply were each drawn up With reference to rms maXimizing
PfOt. well known integrability conditions would guarantee this equivalence 0n the
consumers side there need be no such equivalence. and if we consrder a consumer whose
l purchases balance his total sale of productive set-Vices,such an equality for every
instructive to consider rst, however, the symmetrical case (such
as characterizes markets made up exclusively of entrepreneurs);
and let the speeds of adjustments, HJ, be set equal to one. The
solution of equations (43) can be written
Pi) = Pi + lki". (45)
where (M, ~. -, >)are latent roots of the characteristic equation

and
ana _ X 0120 ... alno
a a a o_ A a un
f0) = .2.
........ .
............. 2...... (46)
Gnlo an2 ' ' ' anna _ X
la -Ul = loi,-" Mal = 0.
the ks depend upon the matrix a and upon the initial conditions.15
As before, stability requires RO,)< 0.
By a well-known theorem of Hermitian matrices, in the sym
metrical case all the roots are necessarily real. If the equilibrium
is to be stable, they must all be negative. According to a classical
theorem, this is possible if and only if a is the matrix of a negative
denite quadratic form; i.e., only if all principal minors alternate
in sign as follows:
Gif Gif" Gaz
au. 0 au..0
|a.-.-|< 0; . .o . .0
> 0; cf.-" 013 u" < 0.
a: 3 au
Ghia ah,- au"
ijki MD
combination of goods and services would, strictly interpreted. lead t an absurdity; it
would imply expenditure proportionality and, hence, zero consumption of every good and
zero offering of every service! For the general demand or supply function we need nt
expect a canceling off of income e'ects" since individuals usually face rms in consump
tion and factor markets.
" If the roots are not distinct, polynomial terms of the form le. e, ' ' "
appear where (: + 1) is the multiplicity of a repeated root. In any se. the Problem
of stability depends only upon the xs and is unaffected by such multipliers since the
exponential always governs the asymptotic behavior of the solution whenever dampening
does occur.
Any ['3th of the form
l 0 0
0 a)! 1&0
0 au du (48)
a "o a "o a tl.
0: o a. an"
.ax . Gt, on"

ISnecessarily negative sngn But such ratios are precxselyequal
to the change the pace of the ath good thh respect to a umt
Increase 1n ltS net supply when appropriate subsets of other puces
are held constant, so that for this case the stabxhty crntena of Pro
fessor HleS are seen to be correct theorems
Where perfect symmetry lS not present (and m busmess cycle
analyms 1t ISalmost always absent), the Hacks cntena are not at
all necessary condltlons and 10 many cases not sufcient " A
system may possess stablhty of the rst kind even though neither
perfectly nor tmperfectly stable m Hicks's sense I long suspected
that perfect stabllnty IS a sufcrent condmon for stability of the
rst kmd But the conjecture proved to be false Perfect sta
blllty, lnke :mperfect stabtlnty, ISneither a necessary nor sufcxent
A word of warning may be m order concerning the mdnscnmmate use of cube:
pnces or quantmes as mdependent vambles Th 3 leads to contradnctory den tronsof
complementanty m the Interary dnsCussxonon page 44 and the mathemat cal denition:
011page 311 the Inconslstency between which can lead to opposne Sign: Such an
Interchange of Independent variables (as bet ecu the Interary defimtxon of stability on
page 63 and the mathematical condmons on 315 and 325) ls parncularly Important
where nonsymmetncal matnces are concerned Does
a Ln
an dp dp; dx dx,
- <0 > 0 here 7
6? (I_x; dpa d?
d? JP:
Imply
4? d?
dz E
9 < o > 0 ?
dx LP: 2
dx dt
1 answer ls Yes but {le proof ts not sxmple but \ mth Symmetry the pmdlfCl
? MI ){dx Mp ) need not be of post vc sign xf more than two \arnables are Involved
condition.17 In any case it is too strict a condition, while the re
quirement of imperfect stability is not strict enough; only in the
case of symmetry do these limits converge. Why any system
should be expected to possess perfect stability, or why an economist
should be interested in this property is by no means clear. Not
working with an explicit dynamical model, Professor Hicks prob
ably argued by analogy from well-known maximum conditions,
whereby a maximum must hold for arbitrary displacements and
through any transformation of variables. As a result, some vari
ables may be made constants, and with respect to the remaining
arbitrary subsets the deniteness of various quadratic forms must
be insured. On the other hand, in terms of a truly dynamic process
the equilibrium must be stable for arbitrary initial conditions or
displacements and for arbitrary nonsingular transformations of
variables, but not necessarily for arbitrary modications of the
dynamic equations of motion such as are involved in the Hicks
procedure of holding subsets of other prices constant (by violating
or relaxing true dynamical relations). In principle the Hicks pro
cedure is clearly wrong, although in some empirical cases it may be
useful to make the hypothesis that the equilibrium is stable even
without the equilibrating action of some variable which may be
arbitrarily held constant. (In connection with the Keynesian
model later discussed, an example of this is presented.)
To summarize: for every case true necessary and sufcient
stability conditions are that RO,-) < 0. where ,-represents the
The matrix
: 1 0 0
0 e 1 0
0 O e 1
1 +1 1 1+6
for sufciently small e's has all principal minors positive and yet has some roots whose
real parts are negative. This shows that even Hicksian perfect stability does not
guarantee dynamic stability. See P. A. Samuelson, The Relation between Hicksian
Stability and True Dynamic Stability," Economem'ca, XII (1944), 256257. Also, as
Metzler has recently shown, the Hicksian conditions are necessary .(but not sufcient) if
the system is to be stable {or all possible (positive) rates of adjustment in different
markets; and if all off-diagonal terms are non-negative, the Hicks conditions are both
necessary and sufficient for stability. L. A. Metzler, Stability of Multiple Markets:
The Hicks Conditions," Econametrica, XIII (1945), 272292. Foroa proof that the
inverse of the last matrix must consist of all elements of the same Sign, see J. Il/Iosak,
General Eqm'librimn Theory in International Trade (Bloommgton, Indiana: PrinCIpia
Press, 1944), p. 49.
274 FOUNDATIONS OF ECONOMIC ANAL YSIS
latent roots of the matnx a The 18not eqmvalent to the Hucks
condltrons
In an address before the Econometrlc Socxety Professor Lange "
suggested that the speed of reactton might be different m every
market so that Instead of wntmg
p: = 11,9: + (49)
as m (43) we write
p. = Hg. + (50)
where H, = H.'(0) rs the (posmve) coecrent appropnate to the
zth market
Even thls IS not sumently general unless a partxcular corn
modnty classxcatlon 15to be favored and nfour formulatlon ISto
be mvnrxant under lmear transformatlons of commodntles and
prices We have no choxcebut to adrmt that the rate of adjust
ment m one market may depend on the excess demand rn other
markets, so that m matrnt terms
where q 15the column of q.'s a the square matrlx a,, and where H
need not now be a posnt1ve dlagonal matrur It would seem
reasonable at rst to requnreH to be posmve demte, or at least to
be posmve quasn demte as the latter expressnon was dened m
chapter vx But If we apply the contragredxent transformation c
as gwen m equation (47) of chapter V1,
9 ""' Cg, : (:_lg
? = ? ? = C'? (52)
" The followmgIllustrations bear this out The system
1h = 291 + 4P:
P: = " PI + ?:
possesses stability of the rst kmd but rs neither perfectly nor Imperfectly stable
The system
p: = Pa "' Pa
11: = 291 + P:
Is Imperfectly stable but departs ever further from equlhbnum
"O Lange Abstract of The Stabxlty of Economic Ethbnum Ecanomemca
X (1942) 176177Puce Henbddy and Employment (Bloommgton Indzana Pru'lcmla
Press 1944) append.
then it can be veried that
13=Hq+---=c'Hq+---=Ha+~-= (cHac")13+---.(53)
It follows that our restrictions cannot be placed upon H alone,
but only upon the product of H and a; that is, the results depend
upon the statical matrix a and upon the dynamical reaction-speed
matrix H. It is sufcient for stability that in some coordinate
system H be positive quasi-denite and a negative quasi-denite,
but these conditions are by no means necessary.
Before leaving the problem of stable multiple markets, I should
like to sketch the e'ect of the introduction of stocks and its rele
vance to stability of the second kind. Let price fall not when cur
rent supply exceeds current demand, but when existing stocks (ac
cumulated over time from the divergenceof current production and
consumption) exceed an equilibrium amount. Then for proper time
units and equal speeds of adjustment
15= Q: _ j; (gs ' gnldr = Q: + J; ,}?!aif(Pi _ ?:)df + '

or
15= a(?i _ Pi") + ' ' v (54)

j-l
whose solution takes the form
{?.-(t) = Pup + '33 (kijelll + hag48:0; (55)

5-1_
where la.-5- Mia-I= 0, and where for unrepeated roots the k's
and the hsare constant depending upon initial conditions. Clearly
the motion will be explosive and undamped unless w/X;are all pure
imaginary numbers; i.e., unless ,-is real and negative.
If the system is symmetrical, this clearly leads to the same
conditions as those for stability of the rst kind. If not sym
metrical, the substitution of second derivatives everywhere for rst
derivatives (through the hypothesis of dependence upon accumu
lated stocks rather than instantaneous ows) implies more ngld
conditions upon the coefcients to insure stability of the second
kind than were previously required to insure stability of the rst
276 POUNDATIONS 01 LCONOMIC ANALYSIS
kind. This is of course because of the requirement that the roots
be real as well as negative.20
ANALYSIS OF THE KEYNESIAN SYSTEM
Up until now I have considered examples drawn from the eld

of economic theory. The techniques used there are of even more
fruitful applicability to problems of busmess cycles. To illustrate
this I shall analyze in some detail the simple Keynesian model as
outlined in the General Theory. Various writers, such as Meade,
HleS, and Lange, have developed explicitly in mathematical i0rm
the meaning of the Keynesian system The three fundamental
relationships stressed by Keynes are (I) the consumption function
relating consumption (and hence savings-investment) to income,
and for generality to the interest rate as well, (2) the marginal
eiczencyof capital relating net investment to the interest rate and
to the level of income (as of a xed level of capital equipment, xed
for the short period under investigation); (3) the scheduleof liquidity
preference relating the existing amount of money to the interest
rate and the level of income.
Mathematically, these may be written as follows.
C(z,Y)Y+I=a, (56)
F(er)I=Br (57)
LU. Y) = M. (58)
here 1, I, I stand respectively for the interest rate, income, and
investment, C, F, L stand respectively for the consumption func
tion, the marginal-efciency-of-capital schedule, and the schedule
of liquidity preference. M stands for the existing amount ol
money, taken as a parameter; a is a general parameter representing
One could consnder the generalization of the intermediate hypotheses of footnote
12 where pn change depends upon stocks and HOWE.namely,
.=anl:am+("h)]
a o ( ap>0
> 0)
if stable for B > O, a = 0, and also {or B = 0, a > 0. it can perhaps be proved to bi
stable for all Intermediate cases
" J E Meade, A Simplied Model of Mr Keynes' System." RMN of Economy
swam. IV (1937), 93-107. J R H1cks,Mr Keynes and the 'CIaSSICS'.A Suggest
Interpretation," Econometma, V (1937), 147159.Oskar Lange. The Rate of Interest
md the Optimum Propensuy to Consume," Economica. V (1938). 12-32
an upward shift in the propensity-to-consume schedule; similarly
as the parameter ;?increases, the marginal-efciency schedule shifts
upward.
We have three relations to determine the three unknowns in
terms of three parameters, viz.:
i = f(a, B, M).
Y = Y(a,, M). (59)
I = I(a, B,M).
As explained in the rst section of this part, the usefulness of the
Keynesian equilibrium system lies in the light it throws upon the
way our unknowns will change as a result of changes in data.
More specically, what are the signs of
d; a il.
da, da' da'
ii LY 115
615, dB' d'
di dY (il
5% ar m?
Differentiating totally with respect to our parameters and evaluat
ing the resulting linear equations, we nd
_di_LY Q__L_" _FLi-FiLY
dcr A ' docA' da A '
__LY __L_.- =(1CY)L'+CL}"
dB _ A ' dA' (16 A (60)
_1CY-FY _Fi+ci
dM_ A ' dM_ A '
il; __Fy(F.+ Ci) + (1 "' Cy "- Fr)]:
dM A
where
C.- Cyl 1
A= F.- Fr _.1 =LY(F.+C.-)+L.-(1CrFr)- (61)
L.- Ly 0
On the basis of a priori, intuitive, empirical experience the
followingassumptions are usually made:
Cr>0. Fy>0, F.-<0, Ly>0, L.-<0, (62)
278 FOUNDATIONS OF bCONOMIC ANALYSIS
whtle >
C . *2-0
and xsusually assumed 1n modern discussmns to be of minor quan
txtatwe Importance
In order to evaluate our nme denvatlves we must be able to
detennme unambiguously the sxgnsof all numerators as well as the
common denommator, A A consnsts of ve terms, two of whlch
are of posntwe sxg'n,two of negatxve SIgn, and one ambrguous On
the bas:s of deductlve analySISalong stnctly statncal lines nothmg
can be Inferred concermng 1ts sign Moreover, even if the sxgnof
A were determmed, all but four of the mne would be found to have
numerators of mdetermmable Sign
Thls Is a typlcal case If we are to derrve useful theorems ue
must clearly proceed to a consnderatron of a more general dynamlc
system whlch meludes the statlonary Keynesxan analy51sas a spe
ela] case Thls can be done m a vanety of alternatwe ways I
shall consxder two, the rst of which xs based upon a dxerentlal
system and ylelds quite demte results
Case 1 Let us assume as before that the second and thlld
relations of margmal efcncy and llQUldltypreference work them
selves out m so short a tlme that they can be regarded as holdmg
mstantaneously Let us assume, however, that I now represents
Intended Investment, and this magnitude equals savmgs mvest
ment only m ethbnum, 1e , when all the vanables take on sta
tlonary values If. however, because of some change, consumption
(say) should suddenly Increase, national mcome not havmg a
chance to change, actual savmgs 1nvestment would fall short ol
mtended mvestrnent because of Inventory reductlon, etc Con
sequently, Income would tend to use Snmllarlyan excessof actual
savxngs Investment over Intended 1nvestment would tend to make
Income fall Mathematically, this hypothesm may be stated as
follows the rate of change of mcomc zs proportzonal to the dzcrence
between mtended samngsmvestmentand actual savmgs mvestmen!
The dxscusswnhere ISunrelated to the controversy over the equal
Ity of sangs and Investment despite possnbleappearances to the
contrary The supercnal resemblance between my formulatlon
and the Robertsoman Identltres whereby the dlerence between
Investment and savmgs xsthe tlme dlerence of Income should not
m-slead the careful reader
Equations (56), (57), and (58) are replaced by the dynamical
ones:
Y=I-|IY-C(i.Y)-al (63)
0 = F(z', Y) I+B, (64)
0 = L(i, Y) -- M. (65)
The solution of these is of the form:
Y = Y"+ ale,
i = io + 028, (65)
I = I + age,
where
C,- Cy 1 A 1
A0) = Fi Fr - 1 = A + XL.-= 0. (67)
Li LY 0
The equilibrium is stable only if .
A
= f.- < O. (68)
But L.- < 0; therefore,
A < O (69)
unambiguously.
This establishes four theorems: an increased marginal efciency
of capital will (1) raise interest rates and (2) raise income; an
increased propensity to consume will (3) raise interest rates and
(4) raise income. But how will the creation of new money affect
interest rates? This can be answered by considering more string
ent stability conditions. Let us suppose that the interest rate
were kept constant (say) by appropriate central bank action.
This assumption is equivalent to dropping the liquidity preference
equation (65) and treating i as a constant in the remaining equa
tions. If the equilibrium is stable under these conditions, we
must have
1
Fy
1=0=(1CyFy)+>\, (70)
- 1
OI'
>\=(1FY_CY)>O. (71)
This leads to another important theorem: (5) the marginal pro
pensity to consume plus the marginal propensity to invest cannot
280 FOUNDATIONS OF LCONOMIC ANALISIS
exceed mzzty or the 33mm mil be unstable (as of a xed Interest
rate) It also tells us (6) that an merease 111the amount of
money must, cetens parzbus 10er Interest rates
We are left n 1th four ambigumes of szgn Tho of them depend
upon the fact that savmgs may \ary m any dlrcctlon 11threspect
to a change m Interest rates If we assume that normally sawngs
out of a gwen mcome mcrease wnth the mtcrest rate, or. nfthey do
decrease do so not so much as does xmestment, then three more
theorems become true 'un Increase In the amount of money (7)
mcreases Income and (8) Increases nmestment (9) an increase
the margmal cfmency schedule Increases Investment There re
mains a nal amblguous term 'h'lt ISthe effect upon Investment
of an mcreased propensnty to consume? Thus is seen to be essen
txallyambiguous depending upon the quantltatu e strengths of the
llQllldlty preference slopes and the marginal cfcrency slopes As
Income Increases money becomes tight because of the need for
nancmg more transactrons Thls tends to depress Investment
As an oset the Increase m Income tends to Increase Investment
through the margmal propenSIty to lmest thch effect wrllbe
the stronger cannot be decndedon (: pnon grounds
I have prepared a 3 X 3 classncatron Indicating the s1gns 01
the mne terms All but four ham demte sxg'ns Of these four
one ISessentially ambiguous as Indicated by a questxon mark The
remammg three show under question marks thelr normal pre
sumptlve sxgns
z Y I
Increase m propensnty to consume a + + ?
Increase 1n margmal efcxencyof capital B + + :3
Increasem amount of money M - l l
][ v.e take Investment also as an Independent parameter (say through government

amen) we lose equauon (S7) and have for atabnlxtythe oondmon
[CrlM'rUhO
l=Cr-l<0
or that the mrgmal prapeusuy to consumemust be less than one But the: u weaker that
the prenons condmon vuew of the fact that the margmal propermty to mvest I:
assumed pos uve
Case 2. I now turn to a system based on a dierenceequation.
It is founded upon considerations similar to those underlying the
KahnClarkmultiplier block diagrams, and for this reason alone
is worth consideration. In addition, the analytical contrasts be
tween dierential and difference systems is brought out. Revers
ing the order of the previous exposition, let us take investment as
an independent parameter and the interest rate as a constant. Let
consumption be a given function of income during the preceding
period of time:
C; = C(i, II:1)= C(Y(_1). (72)
What properties must this function satisfy if the equilibrium is
to be stable? Income clearly equals consumption plus investment:
Y. = C: + I:. (73)
Recalling that investment is treated as a constant, , and using
the consumption relation, we nd
Y; = CGC:) + , (74)
or, to a rst approximation,
(Y; _ Y0) = Cy(Y.1"' Yo), (75)
where
Y" == C(Y) + (76)
is the equilibrium level of income for investment equal to .
The solution of this difference equation takes the form
1/, = Y + K(Cy) (77)
and is stable only if
lCrl < 1 (78)
or
1 < Cy < 1.23 (79)
While the marginal propensity to consume is usually assumed to

be positive, it need not be so, and still the equilibrium can be a
This inequality is in effect the formal justication of Keynes reply to those
criticizing his fundamental law, that the burden of proof lay upon them to explain
why, if their allegations were correct, the economic system was not hopelessly unstable.
See the passages quoted from a letter of Keynes in E. \V. Gilboy, "The Propensity to
Consume: Reply," Quarterly Journal of Economics, LIII (1939), 634. While funda
mentally correct, Keynes does overlook the possibility of other stabilizers such as J
marginal propensity to invest, interest rate, etc.
stable one. Even if it lies between zero and minus one, it is inter.
esting to observe that the multiplier" is positive since
dY I
71T=iCro>' (80)
but less than unity because of negative "secondary" effects.
Let us now drop the assumption that investment is a datum.
although keeping the interest rate constant. Our dynamic system
is of the form
CG. Yll)_ Y: + Ii = 0. (81]
FG, Y.) I. = 0, (82]
and the equilibrium is stable only if
' CY
ill I'- 1} <1, (83
or
|1Fyl<Cy<|lFyl. (34:
Now if the marginal propensity to invest is less than unity
(1 Fy > O)," this leads to essentially the same stability condi
tions as before. namely, the marginal propensity to consume
plus the marginal propensity to invest must be less than unity
(Cy + Fy < 1). But, and this is paradoxical. if the margina
propensity to invest is sufciently large, i.e., greater than +2, the
marginal propensity to consume may exceed unity, and yet thi
equilibrium will be stable! Moreover, beyond a certain critical
value the larger the marginal propensity to invest. the more stablt
is the system This comes from neglecting the lag between Yand I
If we now conSIder the system in which none of the variable:
are taken as given, namely,
C(i. Yt-l) Y: + I: = 0. .
}:(31.Y) _]: =0, (854
LG}. Y.) M = 0.
" In the marginal-eiCIency relation I have made investment depend upon income
which itself includes investment Other writers notably Lange (op at ), have mad'
depend only upon consumption The result is Indifferent since they can be 5110"
to be equivalent [l', however. it is assumed that dI/dC > O, the marginal propensm
1 "W951.d/JY = (dl/dCV + (dl/dC cannot exceed unity If a one-period 131
'5 Ptit (32). then the sum of the consumption and investment propensities must b
less than one.
stability requires that
A + L.-Cy < 1.
m = lLC" (86)
In what may be termed the normal case, where the marginal

propensity to invest is less than unity, this requires as before that _
A < 0, (87)
and immediately all the eight determinations of sign of Case I
become correct.
In the unusual, but possible, case where
1FY<0<CY<(Fy-1)-%(F.-+C.-), (88)
the equilibrium will be stable, but the signs of our 3 X 3 table now
are as follows:
Y
a
[3
In words, the only theorem which remains true under all circum
stances is that an increase in the amount of money must lower
interest rates if the equilibrium is stable.
This example illustrates the additional complexities which sys
tems based upon di'erence equations involve. Later some of the
reasons for this will be explained.
The examples here adduced serve, I hope, to illustrate the light
which dynamical analysis sheds upon comparative statics. Prob
lems in theory and business cycles of any complexity will almost
surely require similar analytic treatment if useful and meaningful
theorems are to be derived.
CHAPTER X
THE STABILITY OI EQUILIBRIUM LINEAR

AND NONLINEAR SYSTEMS
INTRODUCTION
IN THE PREVIOUScmwrnn xt was pomted out that there exrsts an
lntlmatc formal dependence between comparatrve stattcs and dy
namxcs To my knowledge tlns had not prevxousl} been expl1c1tly
enumeratedm the economzehterature and for lack of a better name
I shall refer to 1t as the Correspondance Pnncple It ISthe purpose
here to probe more deeply Into 1ts analytxcal character, and also to
show xts two way nature not only can the xnvest1gatzonof the dy
namlc stablltty of a system yield fruntful theorems 1n statlcal
analysrs but also known properties of a (comparatwe) statlcal
system can be utnhzed to deme information concernmg the dy
namrc properties of a system
An understandmg of this prmcnple :s all the more Important at
'1 two when pure economlc theory has undergone a revolutzon of
thoughtfrom stattcal to dynamical modes While many earher
foreshadowxngs can be found 1n the hterature, we may date thls
upheaval from the pubhcanon of Ragnar Fnschs Cassel Volume
essay of only a decade ago The resultmg change m outlook can
be compared to that of the transmon from classrcal to quantum
mechamcs And Just as the eld of physncs 1t was \\ ell that the
relattonshnp between the old and the new theornes could be part
claned so m our eld a srmnlar nmesttgatlon seems m order
Before entenng however, upon these unavordably techmcal
matters a fen obzterdzcta concerning the fundamental dIFferences
between stattcal and dynamlcal systems may be m order Con
LCIVCCIbroadly a dynamlcal s3stem might be regarded as any set
of functronal equatlons wlnch together wnth untnal condltlons (In
the most general sense) determtne as solutions certain unLnou ns m
functron of tune Accordmg to thls demtxon trmeless statical
Ragnar Fnsch Propagauon Problems and Impulse Problems Dynarmc 500
n cs Econo: : Essays Honor of Gustav Cassel (London 1933) pp 171205
91h
LINEAR AND NONLINEAR STABILITY 285
systems are simply degenerate special cases in which the functional
equations take on simple forms and determine as solutions func
tions of time which are identically constants. We may, however,
dene a dynamical system more narrowly so that it will not be
regarded as truly dynamic if the functional equations involve only
variables of the same instant of time," containing time, if at all,
only as a parameter.2 This excludes the customary statical sys
tems of the historical" as well as the timeless variety. It is
possible, however, that certain subsets of the solutions of the dy
namical equations are dened by equations which are structurally
identical with those which dene a statical system. (Thus the sta
tionary solution of a time-sequence analysis, say of the multiplier
blockdiagramvariety, may be determined by a formula exactly
like that of a timeless, instantaneous system.) This constitutes a
second possible mutual orientation of statical and dynamical
systems.
From still a third point of view a statical system can be.looked
upon as the limiting case of a heavily damped dynamical system.
Thus any statical equation
f(x) = 0
which admits a unique solution x" can be related to a dynamical
system of the form
f{x, + A(x, x")} = 0.
This yields directly the equivalent linear differenceequation
xz+A(xt _xo) :xo,
Of
xt+1
Thus, whatever the initial magnitude of x, at the next instant

the system always takes on its correct statical value. This general
izes easily to systems of more than one variable. .
In the following I deal with interrelationships between statics
and dynamics which largely fall under the second of the three
headings here discussed.
Ragnar Frisch, On the Notion of Equilibrium and Disequilibrium," Reviewof
Economic Studies, III, February, 1936, 100-105.
In chapter xi I deal at some length with the distinction between complete causal.
systems and historical or incomplete causal ones, also with the closely related topic
the generalization of the notion of stationary equilibrium to systems involving .
explicitly.
FUNCTIONAL EQUATIONS AND STATIONARY SOLUTIONS

Startlng out wrth n functtonal equatrons Wl'llChconstitute a dy.
nanucal system mvolvmg 7: unknown functions lx;(:), , x,,(m,
but not tlme exphcxtly,
mew. .xgrnao (:=-1. .n) (1)

we dene a statxonary solution (x1", , :::) as one for which
l l l
Fleo, x200 : xol E O (2)
o -o o-co
These last equations correspond to a set of ordmary statlcal func

trons m n variables (xl, , x,.)
f'(xx. .::..) = 0. (z = 1. .n) (3)
where of course
f(xz. .x..) = 0 (: = l. .n) (4)
The types of functional equations whxchhave been most studied
are those dened by dlfferentlal equations, dnfference equatlons,
and Integral equations and mixed vanetnes The rst of these
possesses the most highly developed theory and provxdes Valuable
examples of various prmcnples Smoe economlc observations can
s1st essentially of senes dened for Integral values of tune, the
second category of dterence equations 15 perhaps of greatest
Interest to the theoretlcal economist
The above classes of functional equations have thxs much 1n
common they can all be wntten as the hmIt of an mnzte set of
equations m an mnnte number of unknowns However, xt xsnot
customary to wnte a system of dl'erentxalequations m the form
(1) , but by use of the smgular Durac &function dened to make the
followmg Identlty formally true,
f(x) = j: so: x)f(a)da. (5)

any lmear dlfferentxalequatlon can be written as an Integral equa
tton Stmxlarlyany mtegral equation of the form
B(t) + f k(a)B(t _-a)da = o, (6)

where B is analytic and k possesses nite moments of all orders,
can be written as a differential equation of innite order, namely,
Bm+wm=a m
where
Ci _ ( D]: 13! k(a)a"da.

_.
(8)
Difference equations and mixed types can also be regarded as
differential equations of innite order; by the use of the Dirac func
tion or by extension of the denition of integration, they can
equally well be represented as integral equations. In the following
treatment I shall investigate formal identities, being little con
cerned with problems of convergence, and omitting rigorous proofs.
There is ample precedent and pragmatic justication for this pro
cedure in all applied sciences.
The greatest attention will be paid to systems of differential
equations and systems of dierence equations. Without loss of
generality these can be written in the following normal form:
dx.
_d=fi(xh''yxn): (i=1mn) (9)
and
A951) = gilxla)! ' ' '1 3700)]! (10)
or
cc,-(t+ 1) = G{x1(t), - - -, x,.(m = x,- + g. (11)
If not already in this form they can be so transformed by the intro
duction of new variables.
For the stationary solutions,
dx.
=o=fm ma a=L ww 0%
or
Ax;=0=g"(x,,---,x,.) =Gx.-. (11=1,---,n) (13)
Systems of Volterra linear integral equations with Poisson
kernels of the form
mw=w+J:mw-mwa G=L~uw(M)
283 fOUNDATIONS OF ECONOMIC ANALISH'
are also of mterest Unfortunately the theory of nonlinear
tegral equatxons xsonly fragmentary
men mm Newman SYSTEMS
Up untnl now most economxsts have concerned themselves wnth
lmcar systems not because of any behef that the facts were so
snnple but rather because of the mathematzcal difculties Involved
m nonlinear systems Thls rs understandable and excusable smce
all thought advances m small steps Nevertheless from the stand
poxntof a study of mdustnal uctuations this can be a rather sen
outs llmrtatlon Thus m a linear system the amplltude of uctua
non depends upon the mmal dnsplacement no Intrmszc amphtudc
-asbetween full employment and zero employmentns mvolved
The attempt to Introduce such a xed amphtude m a lmear system
by the devrce ol determmmg coefcxents so that there Will be
nentherdampened nor exploswe solutrons appears to be nusdzrected
As pomted out In the prevnous chapter page 268 the stochastxcal
dispersmn of the system Increases 1ndemtely Related to this 15
the fact that thusdoes not yield a umque amplltude but one whlch
depends lxnearly upon mmal condrtnons
If we msnst that a system be hnear and that It do not Involve
tlme exphcntly then {or dlfferentxal and dxfference systems we are
restncted to the case ol constant coefczents Thxs type 15mathe
matlcally sxmple and exact solutnonsare known But a hrgh puce
as paid for thrs snmphcnty In terms of specral assumptlons Wthh
must be made
THE NONLINEAR DIFFERENTIAL EQUATION IN ONE VARIABLE

Nevertheless I shall show that the problem of stablhty of eqm
hbnum 1! not that of macrodynamtc busnness cycle analysxs de
pends formally m an xmportant way upon the solution of lmear
systems Thts may be xllustrated by a snmpledufferentzalequation
m one vanable
%=X=f(X)=Ao+A1X+AzX+ (15)
wheref ISanalytic and expresslble as a power serles This equation
presents no difcultres of solutnon smce by a smgle quadrature 1t
may be solved; namely,
X dX
llg=j; =F(X). (16)
Let us suppose a simple" stationary solution X = X exists
so that
f(X) = 0.
f'(X) # 0_ (17)
Then the transformation of variables
x = X X" (18)
transforms (15) into the form
2 f(X)x
:i'=f(x+X) = Li' =0+a1x+agx2+ ---, (19)
where a constant term is no longer involved, and al does not vanish.
Then we can assert:
THEOREMI:A formal solution of the diferential equation (19) can
bewritten as an innite power series in the solution of the simple linear
system
:i: = (11x, (20)
07
x(t) = imam, op = E::cilaewli, (21)

where a is a constant depending upon initial conditions.
That this holds formally (i.e., without regard to the question of
convergence) can be shown by substitution. Thus,
If:= a; z icigl = 20,12 Ckglklj. (22)

l l l
Expanding out and equating coefcients of the same degree in gl,

we nd that each 6 can be determined in succession from the as
E. Picard, Trait d'Analyse, III, 185; G. D. Birkho'. Dynamical Systems (New
York, 1927), chapter iii.
and all prevmus c's
: = arbltrary,
Cz="g
. (23)
C. = M(Cnx. . Cl)
Equation (15) before transformatlon to the form not xnvolvmg

a constant term can m a sumlar iashron be formally solved by an
mmte power senes 111the solutxon of the equatxon
Z : Go, (24)
01
x = K-goa. a) = >: 1cm: + al' (25)

But thlS rs only the conventional power-senes solutlon m : and tells
us nothing about the stabrhty ol the system, every term taken by
Itself goes to mmty regardless of the as, although the mmte
sum need not
It 15well known that such a serres as that m the exponentxals
ts convergent for absolute values of a sufciently small 111a txme
mterval sufciently small But if the system possesses rst-order
stablhty,s 1e , if a; < 0, then the serles wrll converge for all values
of : between to and + ao smce |e*[ ISdecreasmg thh txme, ap
proachmg zero the lumt Smce each term approaches thxs
lzmlt, then
51mx(t, ) = 0 (26)
Therefore, 2f the system possessesrst order stabzlzty, zt necessanly
possesses stabzhty m the small It IS to be emphasnzed that thxs xs
an exactsolntmn : whmhno terms are considered to be a! an order
of smallness and xgnorable Lllcewuse,1fthe system possesses rst
order lustabxhty, It must be unstable m the small However, the
system may be m neutral rst order equllnbrxum.and possess either
stability or Instabllxty 111the small In thrs case a; vamshes, and
we must con51derthe stabnhty of the nonvanlshmg term of lowest
Not to be confused wnth stabnllty ol the rst kmd a concept employed In the
nous chapter
degree. This will be done later. We may state our results as
follows:
THEOREMII: (a) First-order stability is a snicient condition for
stability in the small. (b) The absence of rstorderinstability is a
necessary condition for stability in the small.
EXAMPLE: LOGISTIC LAW

We may illustrate the above principles by considering the
simplest nonlinear system '
X = A0+A1X+A2X2, (27)
where
A12 4A0A2 > 0, (28)
if there are to be any simple" stationary solutions.
Without essential loss of generality, we may by linear trans
formation of X involving only translation and. scale change, and
by scale change in t, bring the above system into the form
a:= x(1 x). (29)
The equation
x x2 = (30)
has roots 0 and 1, each of which represents a stationary state. The
above differential equation will be recognized as one which is satis
ed by the Verhulst-Pearl-Reed logistic law, according to which
percentage changes in a variable fall offlinearly with the magnitude
of that variable, approaching a limit asymptotically. The above
equation, however, is slightly more general since it admits of solu
tions which are not Sshaped.
By quadrature its general solution is found to be of the form
1
x=egy (31)
where K is a parameter determined by the initial value of x at time
zero according to the formula
K _. _1__ _ 1. (32)
_ x(0)
For K = 0 we have the stationary solution
x(t) E 1, (33)
292 F0 UNDATIONS OF ECONOMIC ANALYSIS
and for K = we have the statxonary solutron
xe) % o (34)
It can be easnlyvened that the latter stationary level :s un
stable, while the statlonary level of umty rs approached asymp.
tottcally by all adjacent motlons We may classxfy all posmble
mltml conditions as follows
+lx(0)._"E, 1<K0, } __
o<x(0)-==1. ogK<+ oo. '.LE"() 1' (35
x(0) :O, K: oo, xEEO, )
oo<x(0)<0 -oo<I&<l, lrmx(t)=oo
Let us now apply our expansxon theorem to thns problem
Expandmg around the zero cqmllbrlum pomt. and determlntng the
c coeiczents we readlly nd
x = ae - {ael + {ad}:
lael + = (~1)1[ae'} (36)
1
For gwen values of : and sufcmntly small values of a thts ISreadtly

seen to be a convergent senes equal to the geometric series expan
snon of
as
_ : __1
I (37)
I + 06 1 + et
But for large values of t
WI > I. (38)
and the series dwerges This conrms our expectation that the
zero equnltbnum level Is unstable
The translormatxon
y = x -l (39)
enables us to apply our expansmn theorem to the deterrmnatlon of
the Stability of the other equlhbrxum level Our dnerentnalequa
tron becomes
y a: _ y _ y,: (40)
and m the new varlable the ethbrnum level equals zero De
'Cf A J Lotta Elemausofphysmwwzogymalumm.1925) pp 6468
termining the 5s by substitution as before, we nd
ya) = {Kc-1 + {Ice-12 - {Kc-13 + [Xe-14 --- (41)
= i ( l){Ke}"
1
This will be recognized as the formal expansion of

1
y=W1- (42)
For small values of K and nonnegative values of !, this converges
uniformly, and each term goes to zero as time becomes innite.
Thus, the equilibrium is stable in the small.
It I
: Xlt) l+KQ
I
, 2.0
I
l
: -l.5
I "VK
K-O II
I
___
4K4
I 5//_
I
I
I0
.
K=
' L
. . : l + : .
-4 ..3 ..2 -I O I 2 3 4
FIGURE 4
This example also throws light on the domain of convergence of

the series of exponentiels. It is easy to see from the exact solution
of (31) that the stationary level (y = 0, x = 1) is stable for all posn
tive displacements from equilibrium, i.e., y(0) > 0, 25(0)> 1, and
also for all negative displacements which are less than unity
n absolute value ( 1 < y(0) < O, 0 < x(0) < 1)- BUt the
series expansion of (41) is convergent only for IK I < 1, 01for
(_ %< y(0) < + .% < x(()) < ao) " Thus. its domain of con
vergence xssmaller than the region of true stabnhty It prowdes
only a lower hmxt whnchusnot generally at the same time an upper
llmxt to the reg10nof stablhty
ln Flgure 4 IS depleted the solution of thus equation for all
possxble m1t1al condltlons x(0) The statlonary level of umty xs
seen to be stable, that of zero unstable The diagram bnngs out
one feature not yet mentloned The lower and upper branches
each approach an asymptote for mte values of : Thls means
that for negatxve dlsplacements around the zero level the system
recedes from equlhbnum at an mmte rate of speed after some
mte tune has elapsed For the upper branch It may be loosely
sendthat the system approaches cqulhbnum after commg m from
mmty at an mmte velomty "
Thts example also suggests what ISdoubtless a vahd separa
tion theorem Pomts of stable equzlzbrmm (m the small) are
separated by pomts of demlely unstable egmltbnum, and wee versa
(In the 1nterpretatlonof this, stable or unstable ethbrlum may
be of hxgher order so long as posntxons of one Slded stabxhty
Instablhty are 1gnored )
THE PROBLEM OF HIGHER ORDER STABILITX

Thus far I have conSIderedonly ample" statlonary states, ! e ,
those w1th power serres expressxons m whxch the rst degree term
does not vamsh I now turn to degenerate" statlonary states
those corresponding to multxple roots of the equatxon
X = 0 = f(X). (43)
where
d' X
%:)H) (2.=0, .nlmZ) (44)
In the nexghborhood of such a root the dnfferentxalequation takes
the form
:: = aux + an+1x"+ + , (45)
The expansnonbreaks down at the level at whxehthe curve has an mectlon pom!
! e where
170)= 0 = f Milly
I venture the conJecture completely unven ed that thxs may be a general phenomenon
where
n '; 2.
The solution of the equation
:i: = aux", (46)
where higher powers are neglected, can be expected by analogy
with the previous case to dominate the remaining terms for suf
ciently small initial displacements of equilibrium. In the investi
gation of simple systems this was analytically deduced (viz.,
Theorems I and II); for higher-order systems this conjecture has
not been veried, but is doubtless true. We wish to investigate,
therefore, the stability of solutions of equation (46). If they are
all stable (unstable), we shall say that the equilibrium position in
question possesses nth-order stability (instability).
By elementary methods of integration we can nd the exact
solution to (46); thus
x"dx = andt,
(47)
xln = (1 _ n)ant + x(0)1n_
To solve explicitly for x two cases arise depending upon whether
n is odd or even.
1
x(t) = {a" (1 _ n)t +x(0)1n11/<n1) forneven; (4 8)

1
where the appropriate sign is to be taken so that x satises the

initial condition x = x(0). Both cases may be subsumed under
the heading
1
x(t) = {sgn x(0)ln {an(l _ n + x(0)1n]1/(n1)' (49)
Clearly if n is even, (n 1) is odd. Therefore, the second
term within the brackets takes on either sign depending upon
whether the initial displacement is plus or minus. Consequently,
regardless of the sign of 0for some value of t, and for some dis
placement, the denominator will vanish, which means that x be
comes innite, and the equilibrium is not stable. Actually for 12
even the equilibrium possesses one-sided stability-instability. If
296 FOUNDATIONS 01' ECONOMIC ANALISIS
x(0) nsof the same sugn as a,. the motron wall be unstable. 1fx(0)
:s of opp051te Sign to a. It \\ :ll be stable Thus we haxe oncszded
stablhty instability
If n ISodd then the second term of the denominator ISalways
posmve If and only If a,. ns negative (so that the rst term IS
pos1t1ve)wnllthe denommator stay of the same Sign and approach
znmty as t approaches Innity Hence, no have the following
THEORnn [II If therst nonvamshmg coeiczent1sodd and nega
tue, the system zsstable m the small, ef therst nonvamshmg caecmzt
es add and poszttte, the system es unstable If the rst nonvamshmg
term as even the :) stem possesses one szded stability mstabrhty
Tlns presents a strong analogy to the necessary second) con
dmons for a maumum Write the dz'erentxalcquat1on
X = x) = F(X). (50)
where
F(X) = IXKXMX
Then the lollowmg sums up the results achlcved
THEOREMIV (a) If F(X) a'ords a relatzte marzmum to F.
then X" as a stationary solutton of the dz'erentzal equatzon and
possesses stabzlzty m the small, and comersely
(b) If F(X) a'ords a relatne mzmmum to F then X 1s an
unstable egmltbrmm level
(6) If X as a stationary talue of PM ) uhtch es not an ettremum,
then the system possesses one szded stabzlzty mstabzlzty Alterna
twely af F assumes a statzonary talue and F = f assumes an er
tremumwine, the ethbmcm :: stable unstable
(d) If F'(X ) tanzshestdenttcally.the ethbmcm zs neutral
Thls possnbllatyof lmkmg up the problem of stabxhty thh a
statical maxnmum problem [5 but one spccml aspect of the Corre
spandence Prmczple and one to h1ch ue shall have occasxon to
reier again
AN EXAMPLE OF OM: SIDED STABILITY-INSTABILIT}
VIALTHUSIANAND Oprmuu POPULATION THEORIES
The meamng of one Sided stabxllty Instability may not be
tmtnvely obvmus fortunately a Simple well known economlc ex
ple may be used to Illustrate at Accordmg to Malthus popula
LINEAR AND NONLINEAR STABILITI' 297
tion would increase, decrease, or remain stationary depending upon
the per capita level of subsistence (real income, food, etc.). Let
X = total population, S = per capita real income. Then the
percentage rate of growth of the population is an increasing func
tion of the level of subsistence, passing from negative values to
positive values at some minimum level of subsistence," 5.
Mathematically,
% = MS). (51)
where
MS) > 0. M5) = 0
But the level of income (production) depends itself upon the
level of population (labor) as of given capital, land, and technology.
Moreover, Malthus implicitly and explicitly assumed the law of
diminishing (per capita) returns. Thus,
5 = (X), '(X) < 0- (52)
This last relationship enables us to eliminate S as a variable and to
express the rate of growth of the population in terms of itself.
% = mm; = f(X), (53>

where
f = sb'qs< 0.
and a stationary level X" corresponds to
f(X) = (S) = 0- (54)
The equilibrium is stable because f(X) < 0. If the popula
tion exceeds the equilibrium level, each family will receive less than
the subsistence level, and the population will decline. If it falls
below the equilibrium level (through war, etc.). the average income
will be high, and the population will increase. In Figure 5 the
minimum level of subsistence is shown by MM. Below this level
population will decrease as shown by the long arrow; above this
level it will increase as shown by the other long arrow. R'R is the
returns curve, and the intersection El represents a position of
stable equilibrium as shown by the short directed arrows.
More modern theory suggests the possibility that there may be
increasing returns in the early stages. In this case there may be
two mtersecttons between the returns curve and the mxmmumsub
srstence level, yneldmg two pomts of equxhbrxum In the dlagram
the new returns curve ls R"R, and the new intersectxon pomt ISE,
The new equilibrium pomt 1sunstable smce here
f ' = WW > 0 (55)
If population falls below th1s level It WI" become extmct smce as
returns fall o so does populatxon etc , etc
I .
u
:| 1L
t ll
mu
can:
't!
REM.
r LE
tom. retour On
FIGURE 5
About a decade ago the theory of an optimum populatton

achteved a certam vogue According to one form of thus theory,
at some mtermedzate pomt average returns would be at a mam
mum by educatlon one could raise the minimum standard of
comfort msxsted upon by all famnhes to thxs maxunum level, then
populatton would reach thxs optnmum equlhbnum level Wxthout
entermg uPon the ments or demerits of the scheme, I should like
to pomt out that such an equllnbnum level possesses one sxded
stabrhty Instablhty For displacements from equlllbrlum towards
a larger Populatlon 1t IS stable smce such a movement lowers re
turns and causespOpulatnonto decreasetowards ethbrmm But
(? negatwe dxsplacements of population xt 1sunstable, sancethese
also lower returns and cause populatlon to decrease strll more unnl
the pomt of extmctxon lSreached In Flgure 5 the minimum level
Of subs1stence (comfort) IS 1rrever51bly pushed up by education
exactly to NN ' so that Its Intersectlon, E; wnth R"R represents a
maxxmum of per caplta real Income The arrows 1nd1cate the
one sxdednature of the ethbnum pomt
Analytically,
f = 0. (56)
and the rst nonvanishing derivative is even and negative. Thus,
f is at an extremum, and Theorem IV, (c), applies. The equi
librium is one-sided.8
SYSTEMS OF EQUATIONS IN n VARIABLES

Denition: The system of n differential equations
5:;= f(x1,---,x,,) = Z axj + Z aiikxixk+ ' ' "
J' J.k
(7:=1)"'rn) (57)
where all summations range from 1 to n, will be said to possess a
simple stationary solution (0, --,0) providing that the matrix
(a,;)possesses n (distinct, nonvanishing) roots (>- - -, )),which
are not connected by a linear commensurability relation of the form
ilk! + 1:2)2+ ' ' ' + z.n>\r|.= O (58)
for any set of integers 151,- - -, i not all zero.

In this case a fundamental set of solutions of the equations in
volving only linear terms
5:.-= Z aiixi (59)
i=1
can be written in the form
ale, age, . - -, ane, (60)
where the NS are the latent roots of the matrix (a,-j)or roots of the
secular equation
ou 012 ' am
021 Gaz ' ' ' aan
D(>\) = : : : =(x1A)(x2A)- - {)x).

(61)
anl an2 ' ' ann
' I need not warn the reader against the lack of realism of the above theories in
light of modern demographic trends. For large parts of Western Europe and North
American where net and gross reproduction rates are low, perhaps no level of real income
can lead to a stationary population. Moreover. these rat may fall with increasing
real income; this, however, is presumably an irreversible effect.
300 FOUNDATIONS OF ECONOIHC ANALYSIS
Then we have the followmg theorem '
THEOREMV A formal solutzon of the set of dzerenttal eguatwns
15promded by a power senes m the solutzons of the rst order aqua
tzons, z e
y = G (me me. me) 62
= >: c Acne)+ E mums " + ( )
: : ':
By formal substztutxon thts can be verxed remembermg that the

noncommensurablllty relationshlp (58) ISsatised Each set of c s
can be determmed m terms of prevmus sets and the known a s
The matrix a may have (pairs of) complex roots corresponding
for linear systems to damped or undamped sme cosme terms The
system wxllbe sa1d to have rst order stabxltty 1fthe real parts of
all roots real or complex are all negative smce thrs wrll Imply
damped motion (exponential or harmomc) of the lmear system
An Important part of the stabnhty problem 15the determmatxon of
necessary or sufc1entcondltlons that all real parts are negatwe
and thus wxllbe dealt thh 1n Mathematlcal Appendnx B
Here I should llke to mentlon the generallzatlon of Theorem II
THEOREMVI (0) Fzrst order stability zs a suicnt condmon for
stabzlzty m the small, (6) the absence of rst order mstabzlzty 13 a
necessary candman for stabzlztym the small
Thxs follows because the series (62) can be shown to converge
{or all values of t and values of a: suicnently small If all the real
parts are negative Smce 1tconverges for hmxted values of a and t.
and stnce all terms are decreasmg m absolute magmtude It can
sequently never ceases to converge
A stattonary equll.brlum posmon xs not sxmple 1f a com
mensurabxhty relationshxp of the form mentioned holds Even If
such a relationship does east, the above theorem ISvalld provzdmg
there are no zero roots If no roots vamsh hnear commensura
blllty rclat10n5 mtroduce mto the mmte power senes terms of
the (mm
?)d \ + +uh+ , (63)
where p,.(t) nsa polynormal If the real parts of all A'sare negative.
the exponentlal wdl dommate and the solut1on ml] stnll be stable
'Cl' Ptrd up at Bnrkho' 01) at
Zero roots or ones whose real parts are zero, i.e., pure imagi
naries, cause greater difculties since stability in the small becomes
dependent upon terms of higher degree. 1 am not aware that this
has been completely analyzed in the mathematical literature except
for special cases. I shall not, therefore, enter upon the problem
except to prove a general theorem relating to those many variable
systems associated with the maximum of some function.
Before doing so I may summarize briey the results achieved
so far as follows: Stability in the small Of a nonlinear system of
dierential equations depends except in singular cases upon the
stability of a linear system. This dependence can be rigorously
dened and does not involve a dubious neglect of the squares of
small quantities, etc.
THE STABILITY OF A STATIONARY POSITION

WHICH IS ALso A MAXIMUM
If (X1, - - - , X,,)yields an isolated relative maximum toatwice

differentiable function F(X ., - - -, X), then it is not hard to show
by the theorem of the mean that
F o o _ -_
...-}: Fi(X1| "',Xn) _01 (1 '1 ,?!) (64)
and, assuming we have an isolated stationary value,
F1(Xlr"':Xn)(Xl _Xlo)
+ F2(X1,---,X,.)(X2 - X?) + < 0 (65)
for values OfX sufciently close to X", but distinct from it.
Suppose we are given a system of dierential equations
% ___fi(Xl, . . ., X") = IP,-(Xl,---,X,.). (66)
Only special differential systems can be so written. Unfortunately

space does not permit a discussion Ofnecessary and sufcnentcondi
tions satised by such special systems.
THEOREM VII: (X 1", - -, X,.) is a stationary solution for the
above system, and it is stable in the small.
Transformmg the equlhbnum pomt to the ongm (0, , 0)I
we have
dx.
dt = Fi(X1 + xl: Xo + x): (67)
and F,x. < 0 for sufcxentlysmall nonzero values of x

l
Multlplymg the rst equatron of (67) by xh the second by x,,
etc , and summmg we nd
n dx; _ i nE = !!
'x_dt{},:2] )jF..<0 (68)
For sufcxently small values of x, the sum of squares ISdecreasmg
Hence as t goes to 1nmty 1t approaches a lnmt wlnch cannot be
different from zero If
11m(:5;2 + x23 + + x,.) = 0, (69)
(a
then
11mxl = 0 l1m x2 = 0, , hm x,. = 0 (70)
lo two lo
Hence. the ethbnum IS stable A relatwe proper mnnlmum

yields demtely unstable equ1l1brmm,while a nonextremum sta
txonary value ylelds stablhty mstabxllty
Thxs theorem while not apphcable to all dt'erentxalequatmns,
xsnevertheless very unportant for econormc systems Wlthm Its
scope 1t ISexceedingly general smce It does not requnre that the f s
be analytic, and 1t covers Slmultaneously the stablllty of rst and
hxgher orders Economxcally xt says that the system 1s always
chmblng a hnlland wnllapproach the top
THE.Dmeemce mourns. m. one. VARIABLE

The problem of dxerennal systems has been analyzed m a fmrly
complete manner, and we must now turn to systems of dn'erence
equations, wluch are perhaps of even greater Importance for eCO
nomxc theory The sxmplest case 15provnded by the general non
hnear dterenceequation one variable
XU+ 1) =le(t)}. (71)
01
AXU) = ng(t)l = f - XC!)- (72)

So simple is this equation that we can indicate its solution
graphically, showing on one diagram all possible types of qualita
tive behavior in the neighborhood of an equilibrium position. In
Figure 6 two functions are plottedone relating the succeeding
a
'..lfx)
.

FIGURE 6
value of X (vertical axis) to the given value of X (horizontal axis);

the other is simply a 45 line.
The solution of our equation for any initial condition is shown
by broken lines running between these two functions in the indi
cated fashion. Any initial value X (t) leads to a new value X (t +
indicated by running up vertically to the curve f {X(t) 3; this
304 FOUNDATIONS OF ECONOMIC ANAL} SIS
value must be transferred to the abscxssa order to denve Its
successor Thts ISeected by movmg horlzontally to the 45 lme.
a vertncal movement yxeldsthe next value, and so the process goes
Statlonary ethbrmm posrtlonsare dened by the Intersectnonof
the ! function and the 45 lme, or analytlcally by roots of the
equation
mX=o (m
The diagram Illustrates what are essentnally all possxbletypes
of ethbrra Twenty four pomts of ethbnum are mdrcated
but only smteen represent quahtatxvely dlerent lands of equr
hbrxum, elght bemg duphcates Pomt A represents a posmon of
equlhbnum stable m the small Dlsplacement m e1ther dlrectlon
results m an asymptotrc return to equxllbrmm Pomt B repre
sents unstable equilibrium dlsplaoement 1s followed by an ever
mereasmg one way disequnhbratmg movement C represents a
posxtlonof stable equ1hbnum drenng from A m that It has rst
order neutrality and has only high order stablhty D possesses
rst order neutrality but high order mstablhty therefore xt 1s
unstable 1n the small, and the recessxon from equlllbnum xs mo
notomc E has rst order neutrality but 1ts next nonvamshmg
derwatwe ns exen Consequently, xt has one snded stabnhty
mstabxhty
Thus far \\6 have met only equlllbna ltke those of the one
vanable dl'erentlal system That the smgle d1fferenceequation ls
ncher types of equnllbna xs xllustrated by the variety stnll to
come F possesses rst order stablhty and hence ISstable the
small Unhke A the approach to equlhbnum rs not monotonic,
but rather by means of damped osc11lat10nsof penod two G
represents sxmply a posmon of neutral equnhbrlum m the small.
possessmg rst and high-order neutrality The system stays
wherex er 1t xsput H dlers from F only m that the equ1l1br1um
:s unstable, the divergence from eqmbnum bemg along exploswe
oscnllatlons, I IS snmply a duphcatlon of B, as are K, AI, 0. Q. S.
U, and W
] represents a posmon of neutral ethbrmm The system
oscxllatesmth constant amplitude around tlus posmon (In phys
ICSthis would be called stable equnhbrmm In the termmology of
the prevxous chapter, It has stabxlntyof the second kindnot to be
confused with second- or high-order stability.) L like ] possesses
rst-order neutrality, but unlike . has high-order stability. Equi
librium is approached by damped oscillations. N also has rst
order neutrality, but possesses high-order instability of an oscilla
tory nature. P and R like L and N possess high-order, oscillatory
stability and instability, respectively, but analytically differslightly.
T possesses oscillatory neutrality of all orders, and hence is neutral
in the small. It can be thought of as including J as a special case.
Finally, V and X, while they possess rst-order stability, are
analytically of singular type and should be differentiated. The
former has monotonic stability, the latter oscillatory stability.
The following classication may clarify the possible types.
Nonsingular A
Stable First-order{ Singular V

High-order C
First-order B
Monctonic< Unstable
High-order D
Stable-unstable E
.Neutral G
Nonsingular F
First-order{ Sin ular X

High-order
Stable {Tye
Type 21 L
P
Oscillatory First-order H
Type 1 N
Unstable{ High-order
Type 2 R
LNeutral J, T
We dzscrxmmateanalytlcally between the szxteen cases depend
mg upon the rst and higher denvatrves of f(X ) The rst sub
dlvrsron between monotonie and oscrllatory hmges upon whether
f(X ) rs posrtrve or negatrve Wnthm the monotomc classxcatnon
we haxe rst order stabnhty 1ff rs less than umty and rst order
mstabahty nlf asgreater than unity If j' equals unity, we have
rst order neutrality and must proceed to hrgher derivatives If
all these denvatlves vamsh we have true neutrality m the small
If the rst nonvamshtng denvatwe 18odd and posrtrve, we have
high order stabllrty. llodd and negatwe, ue have high order m
stabrhty If the rst nonvamshtng higher derrvatne as even, we
have one srded stabrhty 1nstab1hty
W1thm the oscrllatory classrcatlon things are even more com
phcated If )" xs less than umty m absolute value, we have rst
order stablhty, ll greater than unlty m absolute value, we hate
rst order Instabllrty ll' f = -1. then we must proceed to
higher denvatxves Should all of these vanrsh we have neutral
osc111atoryequthbnum (1) More generally, If all odd hlgher
denvatwes vanish, the equlhbnum IS neutral and oscrllatory as
ln T When f = 1 and the rst nonvamshmg denvattve 15
odd and posmve, the ethbnum possesseshigh order oscxllatory
stabllrty ll odd and negatwe, the equlhbrxum possesses hnghorder
oscxllatoryxnstabxhty When]" = - Land the rst nomamshmg
derlvattve ISeven then we go on to consrder the rst nonvamshmg
odd derxvatlve As m the prevnous cases, we have stablhty or
mstabnhty depending upon whether thrs ISnegatxve or posntlve
Fmally, we are confronted thh the smgular cases whenf' = O
Il the next nonvantshmg denvatwe rs odd and posture, we have
monotomc stabrhty. ll odd and negative, or If even and of any sagn,
we have stable oscnllatory eqmlrbrnum If all denvatrves vamsh,
the equthbrmm ISperfectly stable as mdxcated m the rst sectton of
thls chapteravhen drsplaced the system returns mstantly ' to equx
hbnum and does not merely approach equrhbnum asymptotically
A few remarks may be m order concernmg the qualttatlve be
havwr of a rstordersystem from any 1n1t1alcondltnon It could
no doubt be shown that 1t must do one of the following (a) go off
to mmty. (b) approach an equlhbnum level, or (c) approach 1
penodlc motion of some mte penod If lt :s revers1ble, that IS
lf f(X) ISnot only smgle valued but adrmts of a smgle valued m
---_.-- _--. -uv-.u4vuan4\ U (11.11.1415
1 du,
verse, then the only periodic motion possible under heading (c) is
one of period two. It is not appr0priate to investigate here the
meaning of the stability of motions more general than stationary
equilibrium levels. When this is done, it will be found no doubt
that there exist valid separation" theorems concerning the order
ing of stable and unstable periodic motions. The necessity because
of continuity (Rolles theorem, etc.) of duplicating certain equi
librium points in the above diagram provides a hint at such
relationships.
ANALYTIC SOLUTION
As with the differential equation systems an exact solution of

the general nonlinear-analytic difference equation can be indi
cated. Let
x(t + 1) = a1x(t) + a2x(t)2 + a;.;x(t)3+ - - -, (74)
or
Ax(t) = (a1 1)x(t) + 0.235(1)2+ a,3x(t)3 + - - -. (75)
We rule out cases where the rst-degree term vanishes in either of

these expressions; and we rule out all cases of singular and neutral
rst-order equilibrium so that al # 1, 0, or 1.
A formal solution of this is given by a power series in the solu
tion of the simpler linear equation
x(t + 1) = a1x(t),
Ax(t) = (al -1)x(t). (76)
Let
gl) = (101. (77)
THEOREMVIII: A formal solution to (74) is provided by

x(t) = caa} +62{aal}2+63laail3+ (78)
This may be veried by formal substitution provided that al # 1,0,
or 1. Each 6 is obtainable in terms of all prev10us 6s and the
known as. .
More generally, we are given n difference equations m normal
form
av.-(t+ 1) = Z ataxia) + deation) + ' ' ' (79)
where the latent roots of a. (M. . )),are never equal m absolute
mluc to zero or one. and where there exnst no hnear commensura
bxlrty relations et the form
m; log A; + 12log l\, + + m,. log A. = 0, (30]
for ms Integers not all vamshmg Then
THEOREMIX A formal solutzan to the system of deference equa
trons es gwen by an ascendmg power smc: m the solutzons of the
lmear system
: + 1) : Z 0.1961(3). (81:
or
x,(t) ==6(a1h1, , a,.)xu) (82
= z:c.,zaM
J + , comm law: +
Each set of cs can be determined by formal substxtutron from al
prewous sets and the known as If some of the roots are multiple
or nf a lmear commensurabthty relatton of type (80) does hold
there wxllprobably exrst a srmnlar power-series solution augmentez
by terms With polynomxal multlphers ln t, provxded always that th<
absolute value of all roots equals nelther zero nor one
All ol the remarks of earher sectxonsconcermng convergence 0
such series apply We must remember, however. the rst orde:
stability of a dt'erence equatron system lmplxes
IX.! < 1. (83
and conversely The followmg theorem ls easxly derived alon1
now famlhar Imes
THEOREMX For a deference equatwn system rst order stabzht;
as a suczent eondmon for stabzlzty m the small, and the absence q
rst order mstabzlzty es a necessary condman
Space does not permnt me to Illustrate the above remarks wrtl
an economlc example. such as would be provxded by the famnha
cobweb theorem apphed to nonlinear supply and demand curve:
OTHER FUNCTIONAL EQUATIONS
\thle this has not been vened, one can hazard the COHJQCtUl
that more general nonlinear functional equations can m the nengh
borhood of an equilibrium point be expressed as a power series in
solutions of the simpler linear systems. Thus, under suitable as
sumptions functional equations can be written in the Taylor-like
expansion
;) + Ffthr), ---, Xn'e):
= {M) Xx}+ z f Km - mxm deT (84)
{Xa(72) _ Xgo}dTIdT2 + - . .,
where K.,, K represent rst and second functional derivatives
respectively. The linear system
X,(t) X. + z f; K T){X,(r) -X,ldr = o (85)

is known to have solutions of the form
X.(t) X: = a,,e*f, (86)

where the Asare innite roots of the transcendental equation
Do) = la.,-+ ]: K,,(v)evdv|= o. (87)

In population analysis single integral equations of this type play an
important part, and the solution in terms of an innite number of
exponentials has been called by A. J. Lotka 1the Hertz-Herglotz
solution of these equations. It can perhaps be shown that an
innite power series in the innite solutions of the linear system
will provide a solution for the nonlinear case.
Similarly mixed equations of the type
y'(t) = f U0). y(t - O)} (88)
may be expressible in terms of the Frisch-Holme linear equation
y'(t) = ay(t) + by(t 0)- (89)
1For an excellent bibliography of applications see A. ]. Lotka, A Contribution
to the Theory of Industrial Replacement," Annals of Mathematical Statistics, vol. X
(1939).
As in the preceding integral equation, boundary or initial con
ditions for this equation necessarily involve an arbitrary function
over an interval, and hence the exponential solutions must be in
nite in number so that an arbitrary function can be expanded in
terms of them. (The same would be true of difference equations
not solely dened for integral values of t. Arbitrary periodic func
tions would enter into the solution, and these would be expressible
in the Fourier innite exponential series.) This raises problems of
doubly innite series and cannot be entered upon here.
CHAPTER XI
SOME FUNDAMENTALS OF DYNAMICAL THEORY
IN THE PREVIOUStwo chapters I have examined a number of eco
nomic examples of historical interest drawn from the elds of theory
and business cycles, and have attempted to suggest the importance
of dynamical analysis not only for its own sake, but as an aid in
arriving at useful theorems in the realm of comparative statics.
Now with less use of illustrative economic material, I should like
to examine analytically the formal aspects of this problem, at the
same time indicating possible directions of generalization.
STATICS AND DYNAMICS
Often in the writings of economists the words dynamic and
static" are used as nothing more than synonyms for good and bad,
realistic and unrealistic, simple and complex. We damn another
mans theory by terming it static, and advertise our own by calling
it dynamic. Examples of this are too plentiful to require citation.
Some writers attempt to distinguish between statics and dy
namics by analogy with what they understand to be the relation
ship in theoretical physics. That this is a fruitful and suggestive
line of approach cannot be doubted. But it is too much to suppose
that very many economists have the technical knowledgenecessary
to handle the formal properties of analytical dynamics. Conse
quently, they become bogged down in the search for economic con
cepts corresponding to mass, energy, inertia, momentum, force,
and space. A case in point is Professor Frank Knights otherwise
stimulating essay on Static: and Dynamics.2
It is certainly true, notably in the writings of Marshall,3 that
economists have made use of biological as well as of mechanical
With a few alterations the following section is taken from my article, Dynamics,
Statics, and the Stationary State, Reviewof Economic Statistics, XXV (1943), 58-61.
Chap. vi of The Ethics of Competition (New York, 1935). This is an English
translation of an article in Zeitschrift fr Nationalo'konamiefor 1930.
See the references to statical methodand biologyin the Index to the eighth edition
of the Principles. In none of his writings does Marshall show more than a passing
familiarity, such as might be expected of any intelligent layman, of the biologicalnotions
of his time. Therefore, he could not be expected to have discerned the lasting truths
3n
analogies, in which evolution and organic growth is used as the
antithesis to statical equilibrium analysis. In general the results
seem to have been disappointing; viz., the haziness involved in
Marshall's treatment of decreasing cost. And if one examines the
more exact biological sciences, one looks in vain for any new
weapon, secret or otherwise, for discovering scientic truths If
the bloodstream lS capable of a simple, abstract, rigorous descrip
tion in terms of the usual laws of physical thermodynamics, so much
the better, if not, one must be content with more complicated, un
wieldyexplanations. Indeed, according to the late L ]. Henderson
the very notion of a stable equilibrium, so characteristic of physical
theory, was actually rst observed empirically in connection with
the resistance of the human body to disease, and formulated by
the ancients as the well-known mc medicamx amme
Nor should the problems encountered in the biological eld be
considered necessarily more complex and less subject to simple
formulation than those in the physical eld Few biological
sciences are less exact than meteorology, which must certainly
be included in the physical sciences.5 Here. simple and abstract
theories spun out from a few assumptions are likely to be inferior
to the intuitive hunches of experienced practitioners, but this is
only a reection of the present primitive state of the subject.
New truths are ascertained in the same way as in more advanced
subjects, and it is to be hoped that rule of thumb may be replaced
by more exact and unambiguous formulations
from the fashions of the moment Nevertheless. writing at the tlme that he did. lt wa!
mewtable that he should have been inuenced, if not oonvrnced, by the Spenoenan
doctrines popular at the end of the nineteenth century.
See J. A Schumpeter. The Theory of Economic Development (Cambridge, Mass.
1934.English edition), Preface. p au,for Mill's notions of stature and dynamic: and then
Intellectual engins
Of course, it (an be said that experimentation ts not possible unmeteorology as m
other physxl sciences But what about astronomy, In some ways the most exact of all
m which no expenrnentatton is possrble?
In dtscussmg the limitations of mathematical methods m economics, ProlessOI
Vmer expresses the belief that the biological character of the subject, so to speak. makes
such methods of limited applicability By this I take him to mean that the sublect 15
complexand difcult, not that any fundamentally different methods of Investigation art
required. See Marshall's Economics, the Man and Hts Times," American Economic
RM, XXX! (1941), 223236
Gustav Cassel ln his Fundamental Thought: m Econowncs (New York, 1925)
chap |, considers Economic Dynamics to be a third stage of analysis, iollowmg a Pu
Static Economy and a quaSI-static" Umermly Progresswe Ec0nomy
FUNDAMENTALS OF DYNAMICAL THEORY 313
Leaving aside all analogies with other elds, there has neces
sarily been within the main corpus of economictheory a preoccupa
tion with dynamics, if only implicitly. The Classical economists
from Smith through Mill had theories of the long-run movements
of population and accumulation. ]. B. Clark rigidly separated
the static from the dynamic in his thought.8 (Other examples
could be multiplied.) Clarks celebrated static state and the
circular ow" of Professor Schumpeter raise a rather vexing point
of terminologythe relation of static to stationarynow cleared
up more or less to the satisfaction of everybody by Professor Frisch.
Stationary is a descriptive term characterizing the behavior of
an economic variable over time; it usually implies constancy, but
is occasionally generalized to include behavior periodically repeti
tive over time. Used in this sense, the motion of a dynamical
system may be stationary: e.g., the behavior of a pendulum satis
fying Newton's Laws of Motion, but subject to no disturbance and
hence remaining at rest; or the behavior of national income after
a change in investment has given rise to dWindlingtransient geo
metric progressions of the usual block-diagram" character.
Statical then refers to the form and structure of the postulated
laws determining the behavior of the system. An equilibrium de
ned as the intersection of a pair of curves would be statical.
Ordinarily, it is timeless in that nothing is Speciedconcerning
the duration of the process, but it may very well be dened as
holding over time. A simple statical system as dened above
would also have the property of being stationary; but, as we shall
see in a moment, statical systems can be devised WhiCllare not
stationary over time.
In dening the term dynamical, at least two possibilities suggest
themselves. First, it may be dened as a general term including
statical as a special rather degenerate case. Or, on the other hand.
it may be dened as the totality of all systems which are 710statical.
Much may be said for the rst alternative; the second, however,
brings out some points of controversy in the literature and Willbe
7See L. Robbins, On a Certain Ambiguity in the Conception of Stationary Equi
librium," Economic Journal, XL (1930), 194214. .
J. M. Clark has wished to carry on where his father left off. to construct a dynamics
which would supplement statics. See J. M. Clark, A Preface to Social Economics (New
York, 1936).
314 I'OUNDATIONS OF ECONOMIC ANALISIS
drscussed here Thus decrsron mvoh. es no ponnt of substance Since
only verbal problems of demtron are xnvolved
We may say that a system 1: d; namzcal 1f 2!: behavwr over[me
:5 determmzd by fzmctzana! equatzons m ukzch variables at dzemzt
pomts of lzme are mzolted m an essentzal uay Thls formulation
15to be attrlbuted to Professor Fnsch Specral examples of such
systems are those dened by dz'erence equatrons 1e those
volvmg a variable md 1tslaggcd \ alues Integral equations whxch
the preeedrng values of the variable enter m a contmuous \\ a}
By a hberal :nterpretatzon of the crrcumlocutxon \anable at a
dxfferent ponnt of trme : may brmg dlfferentral equations under
the demtron remember-mgthat dlfferentral coelhenentschancter
xze the behavror of '1 functron the nezghborhood of '1 pomt
Muted types and more general functronals are Included
Attentron ls called to the fact that \arnbles at different pomts
ol tlme must enter mto the problem m an essenhal \\ a) Thus
a system nmolvmg *!rate of production per umt trme xe a time
denvatne may yet be statue-nl Tlns rs because the variable of
Wthh the rate xsthe time dem atne may ha\c no economicsrg
ntcance It can be Interpreted as the cumulated amount of pro
ductlon from the begmmug of trme or from an rmtlal date no
Signicant economic process depends upon thus vanable The ne
cessrty for the present msustcncc may be apprehended 1frt 15rcahzed
that every vanable can be \\ntten as the dcmame of somethmg
namely rts own Integral Moreoxer a system may be pseudo
dynamic the sense that formal mmnpulatxon of xt penmts us to
reduce xt to statrcal form Unless therefore \\e reserve the desrg
nation dynamrcs for systems whlch :01\e economlcally signicant
variables at different pornts of txme m an zrrcmmable \\ ay \\c shall
nd that no nondynamlc systems evrst
Accordnlg to the present demtlon the lustomcal movement 0f
a system may not be dynamrcal If one year the crop 18high bc
cause of favorable \\eather the next year [on and so [orth tl <
system ml! be statrcal e\cn though not statnonary The same 15
true of a system shoe mg contmuous youth or trend If the secular
' Ragnar Fr sch On the Not on of Equ l br um and D sequ l brrum Ra cw 0}
Economx Sludws III (1935 36) 100106
movement is taken as a datum and if the system adapts itself
instantaneously.10
On the other hand, a truly dynamical system may be com
pletely nonhistorical or causal, in the sense that its behavior de
pends only upon its initial conditions and the time whichhas elapsed,
the calendar date not entering into the process. For many pur
poses, it is necessary to work with systems which are both historical
and dynamical. The impact of technological change upon the
economic system is a case in point. Technological change may be
taken as a historical datum, to which the economic system reacts
noninstantaneously or in a dynamic fashion. Another instance
is provided by a business cycle of a regular periodic character,
which results from impressing an oscillatory outside force upon a
mechanism with an intrinsic (damped) period of its own.
We may distinguish, then, four distinct cases made up of all
possible combinations of static-dynamic and historical-causal:
1. Static and stationary
2. Static and historical
3. Dynamic and causal (nonhistorical)
4. Dynamic and historical
Almost all systems can be placed in one of these categories; and,
depending upon the point of view or purpose at hand, the analysis
may be formulated so as to put a given system arbitrarily in one
category rather than another. Thus, if a system is very heavily
damped so that it approaches its equilibrium value extremely
rapidly, its dynamic features may be passed over in order to
simplify the analysis.
Or a system which is causal from a very broad viewpoint may
be regarded as historical if certain movements are taken as unex
plained data for purposes of the argument. (In fact, every his
torical system is to be regarded as an incomplete causal system.)
I conceive Henry Moores moving equilibrium to be of this statical type, although
the movements around the secular trend are dynamic in character. H. L. Moore.
Synthetic Economics (New York, 1929). _
The notion of musation in a closed interdependent system is exceedinglyslippery
and ambiguous. As used here, a system is said to be causal if from an initial configura
tion it determines its own behavior over time. While it is not appropriate to say tht
one subset of variables uses another to move, it is permissible to speak of a change in
a given parameter or datum as causing changes in the system or in its behavxorover time.
To a meteorologlst economlst a bnsmess cycle caused by weather
disturbances and sunspots would constrtute a causal process But
ordmarnly the economist ISwallmg to regard the causatxon as um
lateral and to adOpt a d1v1510nof labor Wthh he does not study
astronomy but consrders [115Job as done when he has pushed
economic analysrs to a noneconomnc cause " However, there
xsnothing sacred about the conventronal boundanes of economics,
nfthe cycle were meteorologrcal m ongm, economxsts would branch
out m that dtrectlon, Just as tn our day a polltrcal theory of scal
pollcy IS necessary 1f one 18 to understand empmcal economic
phenomena
An Important class of phenomena cannot convemently be
brought under the above four categones I refer to dynamlcal
stochastic processes such as that realized 1fa damped pendulum rs
subjected to random' shocks We shall have reason to dlSCLlSS
such processes 1n connectnon nth dynamlcal problems ansmg 1n
the study of the busmess cycle
For a gwen pattern of shocks as determmed by the partlcular
workmgs of chance m the sequence of t1me under dxscussxon,we
have slmply a dynamlc htstoncal system of type 4 above But nf
we con51derthe totallty of all possrble shocks Wthh may be ex
pected to occur If they are regarded as random draws from a xed
universe, it as clear that calendar trme 15not really Involved, but
only the time whlch has elapsed smce the begmnmg of the process
In this sense xt ls hke type 3 rather than type 4, although the word
causal may no longer seem appropnate
It seems best, therefore. to specrfy two more categones
5 Stochastlcal and nonhrstoncal
6 Stochastrcal and htstoncal
The latter occurs when we have a dynanucal system contammg
stochastlcal variables and, where either the structure of the system
vanes m an essentlal way wrth tlme, or where the unrverses charac
tenzmg the random vanables change m an essennal way Withtime
Sunple examples of types 5 and 6 are provrded by
X ll ix: + {hd,
X 1Xs'l'll' {mc}, .
J A Schumpeter Theory of Economu: Development chap :
where in the rst case the random variable 12which may appear at
any instant of time is drawn from an unchanging universe; and
where in the second case the random variable m is drawn from a
universe dened differently at each instant of time."
In his recent book Value and Capital, Professor Hicks has
given an exceedingly simple denition of dynamics: 1 call Eco
nomic Statics those parts of economic theory where we do not
trouble about dating; Economic Dynamics those parts where every
quantity must be dated (p. 115).
In terms of the above six categories this denition is overly
general and insufciently precise. The second category consisting
of historically moving static equilibria would certainly require
dating of the variables; but it would not thereby become dynamic.
My objection is to his denition, not to his practice, for many of
the systems which he analyzes are in the strict sense dynamic.
CAUSAL SYSTEMS
By a complete causally-determinate system I shall mean one be

longing to the third category of the above classication, whose be
havior is determined by initial conditions (in the broadest sense)
in such a way that its behavior depends only upon the time which
has elapsed since the establishment of such initial conditions.
That is to say, the specication of similar given initial conditions
at a later period of time would result in a similar evolution of the
system except at a constantly later time period. Mathematically,
if :: represents initial conditions at time t, our solution takes
" In the above form the initial formulation of stochastic processesseems to be due
to G. U. Yule. See the references to the work of Yule, Slutsky, and Frisch in H. Wold,
A Study in the Analysis of Stationary Time Series (Uppla, Sweden, 1938). However,
the notions of a MarkoflChain, 3 Brownian movement, etc., go far back in the literature
of mathematics. See S. Chandrasekhar, Stochastic Problems in Physics and Astron
omy," Reviews of Modern Physics, XV (1943), 189.
As is noted later, many systems which involve stochastiml processes n neverthe
lessbe described in part by nonstochastiml models of the rst four types discussedabove.
Often probabilities and statistical parameters may themselves be treated as causally
determined cobrdinates. Thus, if it is known that the rst system described in equation
(1) takes 0n the value k at time O,and that the mean and standard deviation of h are
respectively a. b, then the expected (mn value) of X . is given by the solution of a
causal dynamical equation like that of (1) with the stochastic term removed.
]. R. Hicks, Value and Capital (Oxford, 1939).
the form
x = [Et t.::(t)]. (2)
where :: and i stand for a nite or even innite set of variables
Without this restriction our system would be a hzstoncal one
depending in an essential way upon the time at which the initial
conditions are specied, and could only be written
x = f[1,1.(t):| " (3)
Thus, the differential equation
x+ x = 0 (4)
does not contain time explicitly. and its solution is of the form
x0) = 5:t)e""/", (5)
while
:: + x = t (6)
has the solution
x(t) = (t 1) + [:BU") + 1
zojeU-W. (7)
The rst is a complete causal system, the second IS not. For
the rst a specication of the initial conditions (value of :E)and
a knowledge of the time which has elapsed is sufcient to locate the
quantity x, in the case of the second it is necessary as well to know
the histoncal date at which the initial conditions tire preassigned.
" This represents a one parameter. not a two parameter. family of curves A change
in t mtrodus no new solutions, but Simply alters the mttml conditions t(?) which
identify each
Complete causal systems are essentially unchanged by the transformation
t' = : + a,
but become lustonl ones m the mmescale dened by the transformation
t' = Mt). h"() % 0.
although for simple scale changes dlmenSIOnalconstants serve to maintain all essential
tnvananccs Conversely. m specul ses there may exist transformations (of tune)
which convert a historical system into a causal one, VIZ, the historical system
: + :: 0
dz _
becomes after the transformation
= e". where t' = log:
a complete causal system m the new time t , mmely,
dx+ t')0
dz' 3 "
It should perhaps be explicitly mentioned that no metaphysical
signicance attaches to this denition. Not only is an arbitrary
time scale involved, but also any historical system may be regarded,
as we shall later see, as an incomplete causal one. Furthermore,
the elements of our system may very well be probabilities, which
are themselves determinate although the value of certain variates
may be uncertain.
Merely from the denition of a causal system certain interest
ing properties become obvious. If within a nite period of time a
system returns to exactly the same initial conditions from which it
started, its motion must be perfectly periodic, since it must again
do the same, etc.17 Then again, a determinate economic system
cannot on the downswing pass through precisely the same con
guration as it did on the upswing. Some of the relevant variables
(costs, prices, etc.) must present a dierent pattern or else relevant
dynamic aspects of them must (time derivatives, lagged values, etc.).
Let us now consider some denite causal determinate systems.
An investigator interested in only certain aspects may wish to
concentrate upon a particular subset of variables, neglecting their
mutual interdependence with the totality of all. To a rst ap
proximation the excluded ones may be assumed to be arbitrarily
given functions of time. The system studied is essentially an in
complete causal one and as such is indistinguishable from any
other historical one. Thus, given three bodies, one of large mass
compared to the other two and of great distance from them, it is
often convenient to analyze the behavior of the two smaller ones
on the assumption that they are in a gravitational eld changing
with respect to time (as the external body pursues its relatively
independent path); although in fact the three bodies together form
a gravitational eld independent of time.
Which variables will be taken as data, and which as unknowns
to be analyzed, will depend in each case upon the purpose at hand
and upon a diagnosis of the particular interrelations present.
Often the economist takes as data certain traditionally noneco
nomic variables such as technology, tastes, social and institutional
conditions, etc.; although to the students of other disciplines these
are processes to be explained and analyzed, and are not merely
" G. D. Birkhoff and D. C. Lewis, Jr., "Stability in Causal Systems," Philosophy
of Science, II (1935), 304-333.
320 FOUADATIOIVSOF ECOAOUIC ANALISIS
hxstor) " On the other hand one may llmlt the scope of ones
mqmry mthm the economrc sphere Thus, m studymg a hypo
thetlml short 1mentory cycle one mlght take longer n aves as gwen
Then agam, rt rs often necessarv to go beyond the realm of trad:
tronal economic variables m throv.mg hght upon a partncular proc
ess especnally often, for example, Into the polmcal sphere The
\\ orld rarely ts Into the taxonomlc classxcatrons of pedagogues
Among the numerous reasons for electing to regard certam
vanables as mdependent or arbxtrary parameters, one m partlcular
desen es attention m connectlon mth dynamic analysrs Certam
processes are concened to moxe slonly as compared to others
Thus he dlstmgmsh long run tendenczes from shorter run tenden
cxes, and so forth m mmte regressmn The Interplay of these
processes wxllbe gwen m more detarl later
STATIONARX STATES AM) THEIR GENERALIZATIOV
In chapter lx a stationary or ethbnum state of a dynamic
system of n \anables dened by n functional equatrons of the
general form
mfm. ,xir). z] = o (z = 1. .n) (8)

was prov1ded by a set of constants (xf, , x..)for whlch
l !
F(x1, . x,., !) = 0 (9)
Ex en 1f the system 15not a complete causal one, 1e , 1f lt m

\01\ es tune exphcrtly, it may Stlll possess one or more statxonary
equrhbnum levels Thus, the system described m footnote 16.
p 318, xshlstoncal, but has the statzonary equrhbnum level x E 0
H0 e.er. only m exceptlonal cases WI"a hrstoncal system possess
stationary equlhbnum posmons, and as pomted out 1nchapter nx,
even complete causal ones may lack such p051t1ons It lSdesn'able.
therefore, to consrder the manner m n.thh a statronary equnhbrlum
mn be generahzed
If the vanables of our system are dated, and mvolve time
exphcntly,but are not truly dynamic the Fnsch sense of xnvohmg
vanables at dlfferent pomts of tune (derwatrves, Integrals, and
See ] A Schumpeter. The Theory of Economc Development chap 1
FUNDAMENTALS OF DYNAMICAL TIIEORY 321
more complicated functionals), the manner of generalization is
obv1ous. Our historic (but Frisch-static) system is of the form
F(X, t) = F"(x1, -,xt)= 0.19 (i =1, --,n) (10)
A solution of this system
X=Pw )
such that
F[X(t), :] = o (12)
may be termed the moving equilibrium. However, this is almost
trivial since the moving equilibrium is the only possible evolution
of the system.20 Let us turn, therefore, to truly dynamic historical
systems.
For deniteness, let us contemplate the dynamical system as
sumed by H. L. Moore in his pioneering attempt to determine em
pirical supply and demand elasticities. While he adopted a general
equilibrium approach in which the quantity of each good depends
upon all prices, the formal difculties of the problem will be lessened
if we concentrate upon a partial equilibrium market involving only
one good g and its price p.
The demand for the good is a functional relationship between
its quantity, its price, and time,
9! : D(Pt. t)- (13)
Quite inadvertently, and probably for the purely statistical purpose
of deriving the elasticity of supply simultaneously with the demand
elasticity, Moore assumed that the quantity supplied depended
upon time and price at a previous period of time,
q. = S(Pz_1. t). (14)
The two equations together form a dynamical system which
determines the evolution of (p, q) for given initial values of p or q;
" I shall have frequent recourse to matrix notatiOn whereby a set of variables
(xl, - - -, x,.) will be represented by (X), a set of functions [f(x;, - - -,x,.),- - -,f"(x;, - - -,x..)]
by x): etc. Thus, f(x) = 0 impliesf(xx. ' ' .In) = 0, where (i = l, -- , n).
2Quite possibly one might imagine random or chance errors causing deviations
from the equilibrium dened by the equations (10) so that the moving equilibrium repre
sents a smoothed out trend of the actual observations. One might have f(x + e, t) = 0,
where e is a random variable.
322 I'OUNDATIONS OF ECONOMIC ANALlSIS
namely,
P: : fl?- 50:01
m=gBP an
There exrsts a one parameter farmly of motlons, each motnon IS
determmed by efcacrous forces, but not all can be termed movnng
equxlnbna or the term loses srgmcance
If tune did not enter eXpllCltlymto our equatlons, xe , 1fnerther
the demand curve nor the supply curve wereslnftmg, the statlonary
equuhbrlum would clearly be dened by
D(P) _ go = 0,
w=0 (m
1
For thxs steady motlon, and here demng Ap, = p, - p. 1

P: = Ptx= = P",
q: = 9M = = T, (17)
Ap, E Ag. ?- 0
When time ISexpllcrtly mvolved what 15the analogous posntron
of movmg equxllbnum? Let us consrder the followmg alternatwe
demtlons wh1ch have been advanced (1) the demtlon which
Moore seemed to urge whereby the posmon of movxngequlllbnum
15to be represented as a statrstlcal trend. (2) the movnng equn
lrbnum dened by the equality" of supply and demand, (3) the
movmg equlllbrmm dened by the 'Frlsch barnng process to be
descrzbed later
1 The representatlon of equ1|1br1umas snmply a statistlcally
tted trend has apparently no unlversal vahdnty In particular.
xl thxs method IS applled to the causal case (especrally where the
approach to equrlnbnum 15nonosc1llatory), 1t need not lead to the
correct equrhbnum level, nor necessarily to any stationary level
Probably Its apparent acceptance by Moore rested m part upon Its
supposed relation to the cntenon to be descnbed next For
stochastrcal systems, discussed later, trend ttmg 15 more de
fensrble, but stnll not optxmal
2 The criterion that movmg equlhbnum be dened by eq
lzbrtumof supply and demand seems on rst snghta natural general
rzatlon of the statnonary case But, upon exammatlon, ambxguxtles
appear 111this formulation Gwen any pme demand reacts m
stantaneously, supply after one year. What sense is there in
equating the two? (Of course, each year in the shortest run price
is determined by the equality of demand and short run supply for
all possible motions.) Nevertheless, if the shifts are very slow,"
one might as a rst approximation disregard the differences be
tween p; and pc-; and equate
D(P, t) = S(Pt1;t) = 5(1)! ' AP !) S(Ph t) (18)
on the assumption that Apt 0. Solving
D(Ph t) _ SU?! " 0. t) = O. (19)
we get
Pc = Mt), (20)
and this might be termed the moving equilibrium for price. Para
doxically, on the assumption that price is not changing, we derive
a moving equilibrium path for price! But this is characteristic of
the method of successive approximation. Assumptions known to
be in error are made, later to be corrected. A second approxima
tion would be derived by making
AP: = 41910) (21)
and equating
DEMO). t] - SUM!) - A1910).3 = 0 (22)
and the nth approximation by the relation
DEPnU), 1] - SEPnU) APniU):1] = 0- (23)
If the sequence of functions [p.-(t)] converges uniformly to a limit
function p'(t), then 13(t)is a solution of our original system.
The method outlined above is essentially identical with the
moving equilibrium of a biological or chemical system undergoing
slow changes.21 If the system is dened by
i = f(x. t). (24)
where f. is small, the moving equilibrium (rst approximation),
x1(t), is dened by
agi) = O = f(xl, t); xo = constant; (25)

21See A. J. Lotka, Elements of Physical Biology (Baltimore, 1925), chap. xxi. This
contains numerous references.
the second apprommatxon x,(t), by
::1 = foe: t). (26)
and the nth by
z, = f(xn !) (27)
The sequence of functxons [n] may or may not converge um
formly to a hmlt functxon, 1fit does converge the hmxt function 18
one soluuon of the ongmal (Instead of a Single variable any
number could be involved Without a'ectmg the argument )
The above demtlon has the followmg advantages m Its favor
For nonhxstoncal causal systems not mvolvmg tune explicitly, It
does yxeld the correct posmons of stationary equlhbnum at the
rst approxnmatlon Also the successwe approxxmatlons are m
varrant under changes ol vanables of the form
y = f(x) x = f(y)
and changes of tlmc
t' = g0). t = 5%)
Nevertheless, these advantages do not outweigh the overwhelm
mg shortcomings Involved m thus demtxon What after all ls
bemg approxrmated? Suppose all SOlUthllSof equauon (24) were
known exactly to be of the form
x = Mt. z) (28)
There would then be no need for approxlmatxons at all Wh1chof
these solutions could legmmately be termed the movmg equlhb
mum? The partlcular method of successwe approx1matxon out
hned above, If nt converges at all, selects out arbxtrarnly and
mystenously the particular solutlon to be dlgmed by thls txtle"
I reject, therefore, the above method of approxnmatxon as demng
a unique movnngequzhbnum, although It may be approprlate for
amount;
at par sab
For a [meat dnfferentlalequatlon of the form
:: a(t)x = b(i), (29)
the general solutlon equals any particular solution plus the general
"The sums ve lunct ons the approximation sequence x (I) do not m general
sattsly the same mural condmons so that only m the hmrt can one determme which
mutualamdmons the nal solution attamed ml! satxsfy
solution of the reduced homogeneous equation
1i;' a(t)x = O. (30)
Let
x = cu(t) (31)
represent the general solution of (30), where 6 is a parameter
specied by initial conditions. Let 111(1)be a particular solution
of (29). Then
wc) + cu(t) (32)
is the general solution of (29). Clearly also
two) + cum] + kun) (33)
is a general solution since the expression in brackets is a particular
solution for a specic a.
In some elds (electrical network theory, etc.) special signi
cance is attached to that particular solution which contains no
terms of the form [au(t, and it might be thought that this might
be an acceptable denition of moving equilibrium. Where the
functions involved are not the simple elementary ones, the above
criterion does not specify unambiguously a unique function. But
it can often be modied to do so. In fact, the limit function
dened by convergent approximation sequences of the type de
scribed in the previous section often represents this particular solu
tion. I am not aware, however, of any useful purpose to be served
by embracing this denition.23
There is one interesting interpretation of the rst approximation
which deserves detailed attention. One often encounters the no
tion of an equilibrium which, before it is reached, recedes because
of historical perturbing forces. Imagine a boy carrying a Sto? at
the end bf a piece of string. If he were standing still, equilibrium
wouldbe attained when it hung vertically at rest. But as he walks
the stationary equilibrium is never attained, the position in whlch
In linear systems (electrical, etc.) where there is an impressed periodic force, the
general solution of the differential equations can be written as the surn of a purely
PETOdcmotion and a transient. In the mse of damped systems the transrent necessarily
goesto zero in the limit, and the motion necessarily approaches the purely periodic func
tion. For some purposes it might be convenient to dene the purely Pf_d' "tdof
as a moving cQulibrium" and give it privileged treatment. particularly the P C
function is a pure sine wave.
equ1hbr1um might be possrble 1f lus motlon were to stop belng
constantly displaced
We have here what may be termed recedmg equxhbrzum For
the tlme t 1t 15the posmon of statlonary equlhbnum whrch nght
be attamed 1fsubsequent to : all hlstoncal changes were suspended
That ISto say, m all our functronal equations a bar rs placed over
t wherever xtoccurs exphcntly for t1mes subsequent to : The result
ls a hypothetical or vu'tual complete causal system For a new t
we have a corresponding vn'tual equ1hbrzurn and thus 18dened
our receding equlhbrmm as a functlon of tlme
Let us examme a snmple example to Illummate thlS concept
Let
:: + x = t (34)
be a historical system At : equal to ! the receding equlhbnum
asthe statnonaryethbnum correspondmgto the causal system
% + x0) = :. (ss)
where the right hand member IS treated as a constant Clearly,
such a posrtlon ISgwen by
O+x=t (36)
But this lSpreusely the rst approximation x10) of equatlon (25).
Specrcally m thns case
dx
d;=0=zxl (37)
The Virtual path of the system after the freezmg of hlstorlcal
change ISonly one possrble solution For other mntxalcondrnons
at tlme : dl'erent solutlons Willbe dened In partlcular, let us
consxder lmtlal condmons u(i, a:), where u 18a true solutlon of the
lustoncal system (34) The corresponding Virtual path may be
written
:. & v, u(t, x)] (38)
For the specxcsystem dened by (34)
u(t, :) = : -- 1 + (a:+ De, (39)
aft, u(t, z)] = ( c + x + De +! (40)
FUNDAMENTALS OF D YNAMICAL THEOR Y 327
The actual solution of the historical system is seen to be an envelope
to the family of virtual solutions vl], u(i, i)], The virtual move
ment for short times approximates closely the true movement in
that they are mutually tangent. That is, if the boy weresuddenly
to stop walking, the pendulum would change its motion; but the
new motion would be tangential to the old, and for short intervals
they would not diverge much.
The virtual curves approach asymptotes or hypothetical sta
tionary equilibria. These equilibrium values plotted against time
form the curve of receding equilibria. This function of time is not
in general a solution of the true dynamical equations. The sta
bility of the hypothetical equilibrium can, however, be determined,
and, as will be seen later, bears some relation to the stability of the
actual motions.
3. I turn now briey to the normal values or equilibrium
values dened by Professor Frisch. This is dened by relaxing
a subset m of our n functional equations
fitxlr). x,.ig), a: 0, (a = 1. n) (41)
and replacing them by a set of m hypothetical equations

1 !
gi(x11 . ' ') xn: t) = 0- (42)
In the new set of n equations the instantaneous forms of the vari

ables (at least in some of the places where they occur) are barred
and regarded as unknowns. Everywhere else (in all dynamic
forms of the variables and possibly in some places where the'in
stantaneous forms appear) is substituted one solution of the original
set of functional equations denedfor a particular set of initial con
ditions. This yields n historical equations of the special form (10)
" It is possible for the receding equilibrium to be one solution as. for example, "1
the system
+y=h
y = t is the receding equilibrium and at the same time an actual motion of the system.
If the boy has always been walking at a steady pace, the pendulum may be hanging
vertically.
" R. Frisch, On the Notion or Equilibrium and Disequilibrium."Rm of "
nomic Studies, III (1936), 100105.
. u" {"
328 FOUNDATIONS OI ECONOMIC ANALYSIS
in the barred variables, (51, - . i.). This denes the moving
equilibrium.
Professor Frisch is well aware that this denition does not
correspond in the stationary case to the concept of equilibrium; he
even thinks that the former will eventually replace the latter as
the tendency to formulate the economic reasoning in exact dynamic
mathematical terms gains ground. But this concept, he be
lieves, is what many modern writers have in mind.
Whether or not this last conjecture is true, I do not know, par
ticularly since I am unacquainted with the original Scandinavian
writings.27 But for the present purpose this is irrelevant; it is
only important to point out the differences between this concept
of a normal and the moving equilibrium for which Weare looking.
Some of the salient properties of the normal" movement are.
(1) It is in general not a solution of the original set. There
exists no set of initial conditions which would produce it in actuality.
(2) It is not unique, but depends upon as many parameters as
are necessary to specify initial conditions for the original Set of
equations This is because it depends upon the particular solution
out of an innity of posable solutionsinserted in the modied
set of equations f and g. (See the italicized sentence above.)
(3) Being dened by essentially static, albeit possibly historical
equations, there is no sense in asking questions concerning its
stability.
For all these reasons it is unsatisfactory as a representation of
moving equilibrium for our present purposes
[bid , p 102
" Professor Frisch illustrates this concept by referring to the Wicksellian relation
between the actual and natural rate of interest. In so rationalizing the arguments of
some neo-chksellians he is, I fear, over charitable, ascribing to them an undeserved
degree of sophistication Actually, Wieksell himself considered for the most part a
system of impliut dynamic relations not involvmg time explicnly. i e , Without historical
change (Witness. for example. his controversy with Professor D-iindson on the equi
librium requirement of stable prices in a system undergorng output expanSion because
of irreverSible technologin change ) ithin a nonhistoricnl framework it is possible
to construct dynamic models of the Wicltsellian system in which the natural rate of
interest represents the stationary equilibrium level of the system and not the solution
of a hypothetiml alternative system If this be granted, what is the corresponding
condition for a system in which elements of historical change have been introduced?
This brings us back to our original query.
Still another notion of morvmgequilibrium is suggested by an explosive exponential
RESOLUTION OF THE PROBLEM
We have examined in turn alternative denitions of moving

equilibrium and found good reasons for rejecting each one. This
suggests a reexamination Ofthe purpose for which it is desirable to
nd a concept of moving equilibrium.
It will be remembered that in the case Of nonhistorical causal
systems there was found to be no ambiguity in the denition of
a stationary equilibrium state; in fact, it was the hope of generaliz
ing the notion Of a stationary state for historical systems that
motivated the search for moving equilibria.
Let us go back a step and ask why we were interested in a
position of stationary equilibrium. Clearly it is only one out of
an innity of possible motions of the dynamical system in question.
But, and this suggests the answer to our question, if the stationary
equilibrium is stable, all motions approach it in the limit.29
We were concerned then with the behavior of all the motions
Ofthe dynamical system; and it was simply in the nature of a co
solution of a causal system such as
y = 0.
This has solutions of the form
y = ke.
On the above denition every motion but the stationary equilibrium one would be a
moving equilibrium. By denition all the moving equilibria would be unstable. The
effect of small shocls would be multiplied through time. On the other hand, the variable
(/y) would be relatively stable.
The above concept which applies, for example, to a pOpulation with constant specic
fertility and mortality is in contrast to the moving equilibrium represented by the
growth curve of an infant. Given a small shock (measles, etc.), the child's weight will
deviate from normal" growth, but afterwards will catch up so that no evidence of the
interruption will remain in the later progress of the child. Similarly, an economy re
covers from the effect of (say) a war and continues on its secular path. But the popula
tion referred to above is forever dicrent in absolute numbers after a shock (war,
etc.) even though its relative age distribution and other features may again approach
stable form.
Let x(t) = :? represent the equilibrium position of a system. Let :(!) represent
any other motion. Then, if stable,
lim :(!) = x;
l'om
i.e., for any given positive e. however small, there exists a t such that
lx(t) xl < e, for t > 1".
But this is symmetrical in :(!) and x so that the equilibrium position can be said to
approach any motion, as well as vice versa. It easily follows that it"(t), any motion,
approaches :la), a particular motion selected arbitrarily.
33o FOUNDATIONS or ECONOMIC ANALYSIS
1nc1dencethat this problem could be studied by an exammatron of
the properties of a smgle specxalone It wrll perhaps be enlighten
mg to pomt out that even 111the nonhxstoncal case we can Just as
well concentrate upon any other smgle motxon than the equlhbnum
one For 1fthey all approach the equrhbnum motion then they
all approach each other And every motlon must approach any
one selected at random
Thxs suggests the followmg answer to our dllemma We shall
concern ourselves not wzth the stabzhty of a pamcular manon of a
hastortcat system winch can be gwen the prwzleged tztle of momng
egmhbnum but rather unth the stabsltty of each and ezery manon of
the system Thls WI" not preclude our deSIgnatmg for specnal at
tentlon partlcular mot10ns wrth specnalproperties (as for example
stnctly perrodlc motions steady motion m whlch some coordmate
Increases Imearly etc)
Thns suggests how I would answer the purely verbal problem
what processes shall be demgnated as egmlzbrzumprocesses? Once
we have strlpped the equrhbnum concept of normatlve and teleo
logncal connotatlons 1t does not much matter how we apply the
term The term may be applted to stattonary \ alues only From
thus narrow pomt of vrew a competntlve Industry would be m CQUI
hbnum only after all long run condltlons had been satlsed entry
of rms correct output for each rm etc
For other purposes we mlght apply the term to those mter
mediate short run sntuatlons m whlch each rm IS producmg at
pnce equal to margmal cost (and above average vanable cost)
even nf price dxverges from average cost so that the number of
rms ISchangmg or xsabout to change It rmght be reserved for
Situations of the last descnbed type where the further condrtron IS
reahzed that the number of rms rs changmg at a causally deter
mmed rate Stxllothers may be wrllmg to apply It to momentary
condrtlons of supply and demand 1nclud1ngvarlatlons m speculative
stocks of the good m questlon
I, myself nd lt convement to vnsuahzeethbnum processes
of qunte dxfferent speed some very slow compared to others
Withm each long run there ls a shorter run and thhm each shorter
run there ISa stxllshorter run and so forth m an mmte regresswn
For analytlc purposes 1t lSoften convenrent to treat slow processes
as data and concentrate upon the processes of mterest For 6!
ample, in a short run study of the level of investment, income, and
employment, it is often convenient to assume that the stock of
capital is perfectly or sensibly xed. Of course, the stock of capital
from a longer run point of view is simply the cumulation of net
investment, and the reciprocal inuence between capital and the
other variables of the system is worthy of study for its own sake,
both with respect to a hypothetical nal equilibrium and the simple
course of growth of the system over time.
So to speak, we are able by ceteris paribas assumptions to dis
regard the changes in variables subject to motions much slower"
than the ones under consideration; this is nothing but the per
turbation technique of classical mechanics. At the same time we
are able to abstract from the behavior of processes much faster
than the ones under consideration, either by the assumption that
they are rapidly damped and can be supposed to have worked out
their effects, or by inclusion of them in the dynamical equations
(derivatives, dierences, etc.) which determine the behavior of the
system out of equilibrium.
The rst of the above mentioned alternatives constitutes the
justication for the use of comparative statics rather than explicit
dynamics. ~ If one can be sure that the system is stable and strongly
damped, there is no great harm in neglecting to analyze the exact
path from one equilibrium to another, and in taking refuge in a
mutatis mutandz'sassumption. Of course, if one chooses to neglect
certain dynamic processes, one may still retain others; e.g., in
studying capital formation over two decades I' may choose to
neglect inventory uctuations, but still may retain the acceleration
principle in its secular aspects.
Under the second' alternative where shorter run processes are
contained in (say) the differential equations of the system, it is to
be understood that these differential equations do not necessarily
hold exactly at each instant of time. There may well be a still
shorter run theory which explains how still higher differential equa
tions lead to (rapidly) damped approaches to the postulated differ
ential equation relations. And so forth in endless regression.
It may be argued that so general a connotation is at variance
with traditional usage of the word equilibrium. Is it not straining
language to think of a cannon ball as being at equilibrium, not
only after it has fallen to the ground at rest, but also at every point
332 FOUNDATIONSOF sconowc ANALYSIS
m Its lght, when It ISon Its mean trajectory as well as m 1ts pre
cessxonaround this path? Perhaps such termmology may occa
sronallylead to confuswn, however, thh carefully stated quahca
trons It may be convenient
To examme the stability of any motion [u,(t) , m)] sunply
substxtute mto the functional equatlons
x.(t) = 24.0) + m0) (43)
where the above equation ISSimply '1denition for the :7'5 There
result n functnonal equatlons m the r;s smce the u s are gwen func
trons Because the u's constitute a solutlon of the ongmal system
there emsts a solution m the n s of the form (0 0, ,0) The
ongmal solution w1ll be stable nf for all poss:ble :mtlal conditions
hm n (t) = 0 (1 = 1 n) (44)
lou
Moreover provnded the orngmal functional eqmtxons are of a

very general class (differenttal equatnons, difference equations, m
tegral equ'ltlons etc) the resulting equations can {or sufcxently
small displacements be regarded as lmear m the 11's They are not
however Independent of time. but ordinarily wnll mvolye 1t
e\plzc1tly
Thls 15most easnly seen for a smgle vanablc Let our rmplzmt
functional equation be
fCuw i ne) :] = o. (45)

where
fEufr). :] = 0 (46)
because a ts a solutlon Prowdmg certain assumpttons are made

concermng the contmunty of the functional f and Its functional
derwetwee, ei htgbm order, then-Q.ex.asav. 'pmem very mud).
like Taylors expansnon for ordlmry functions of a pomt This
takes the form
ue)ine) :] = 0 + f_m,n)q(n)dn+ , (47)

where K . xs the functional derxvatnve of f This seems to result
only m lmear Integral equatxons, but by use of the Stleltjes Integral
FUNDAMENTALS OF DYNAMIC/1L THEORY 333
or the Dirac operator (x) with the property that
[Qmma=ma mm
and
Emva=na m)
the general linear differential and difference equations can easily
be written as linear integral equations. Thus,
% + x(t) = 0 (50)
takes the form
_ K(t,1)x(r)dr = o, (51)
where
KU, 1') = '('r t) + (1' - t). (52)
The convertibility of the problem of stability at least in its
rst order aspect into an examination of linear systems is of in
estimable value since the largest part of existing mathematical
knowledge relates to such systems.
CONCEPTS OF STABILITY
In chapter ix there was some discussion of various kinds of

stability. Our treatment would not be complete without at least
a cursory survey of the various senses in which this term has
been used.
(a) It has sometimes been used in a very broad sense. Any
position of equilibrium is stable if deviations from it remain
bounded. If no motions go out to innity, then each is stable.
For many purposes this is unsatisfactory. Better names than
stability can be suggested for this property. An egg balanced on its
tip upon a level plain would be stable in the above sense.
(b) Poincar implied still another denition when he coined
the phrase stability in the sense of Poisson. This means that,
although the system may not exactly repeat itself starting from an
arbitrary initial state, yet in general it will return to the vicinity
of its initial state and nearly repeat its motion during a long interval
of time." 3 This is quite different from the usual meaning of
stability, and the term recurrent is perhaps more appropriate to
systems employing this property.
(0) A very common use of stability in nonconservative physical
systems is the one which I have termed stabzlz'tyof the rst lmzd.
It holds when every motion approaches in the limit the position
of equilibrium (and every other motion). It is not reversible in
time; going backwards, all stable systems become unstable.81
Subdivisions under this heading are of course possible. Thus,
there is stability of the rst kind in the small if in a sufciently
small neighborhood of a given motion all motions are stable. First
order stability of therst kmd prevails when certain sufcient condi
tions hold; namely, when the linear terms in the expansion of our
functional equations taken by themselves yield a system which is
perfectly stable We have been concerned almost exclusively with
this denition of stability.
(d) All conservative physical systems are reversible in time, and
volumes are preserved in phase space. This rules out stability of
the rst kind. A conservathe system free oi the dampening dissi
pative force of friction, if displaced from stable equilibrium, never
returns to rest at stationary equilibrium A pendulum displaced
from the perpendicular has higher total energy (equal to potential
energy plus zero initial kinetic energy) than the equilibrium posi
tion since the latter represents a minimum Total energy must
be consened. so, in passing through the equilibrium position, the
system must possess motion in order to prov1de some kinetic
energy to be added to the minimized potential energy.
(e) There remain still other notions such as permanent stability
(the previous denition holding for all time between minus and
plus innity), wma-permanentstability in \\ hich the above proper
ties hold for long periods of time, complete or trigonometric sta
bility in which the motion can be approximated by certain har
monic sums, etc.82
"G D Birkho and I) C Laws, Jr , Stability in Causal Systems," Philosophy
of Scmice, II (1935), 310
This might be termed stabxln) m the sense of Liapouno'. (See E Picard, Trait:
D'AMJM. III. p 200) It is also shown there that instability of the rst order rules
out stability in this sense Birkho has termed this unilateral stability.
See G D Birkhol'f, Dynarmcal Systems (New York, 1927), chap iv, for more
detailed discussion.
FUNDAJIIENTALS OF DYNAMICAL THEORY 335
It is unnecessary to go further into this problem except to point
out that unlike the greater part of the subjects treated here, the
analysis of the concept of stability with the implied investigation
of the qualitative behavior of generalized dynamical paths leads
into some of the most difcult problems in higher mathematics.
NATURE OF THE BUSINESS CYCLE
We have seen that there are many interesting and fruitful as

pects of dynamics which have nothing to do with the business cycle
as such, but which are important for the understanding of processes
usually classied under the heading of economic theory. However,
of all branches of dynamics the one which has received the greatest
attention is that dealing with the uctuations in employment, in
come, and general business activity. Numerous explanations, as
sociated with a variety of names, have been put forward. In fact,
there have been so many different theories that it has been neces
sary to devise a number of different systems of classication in
which to catalogue them (e.g., underconsumption, overinvestment,
etc.; or frictional, institutional, monetary; exogenousversus endo
genous; etc.). I should like to survey these in a brief way in order
to isolate the analytical differences involved rather than to concen
trate on the historical, institutional, and personal variations.
(1) It is a commonplace today to employ the terms exogenous
and endogenous in describing cycle theories. The former refers to
theories which nd the origin of the cycle in some outside, non
economic datum which varies in a quasi-periodic manner, and by
one way causation engenders a cycle in economic time series. A
sunspot or meteorological theory is usually offered as the prototype
of this class. Inasmuch as the economic fluctuations whose ex
planation we are seeking are not themselves strictly periodic, or
more than quasi-oscillatory, it is not really necessary that the
exogenous factors be any more regular in their periodicity and
amplitude. Nor need the exogenous factors be completely inde
pendent of reciprocal inuence from the economic system; strictly
speaking, it is only necessary that lines of causation be largely one
way in direction, from the outside in. Analytically the extreme
exogenous theory is analogous to a forced periodic motion, where
the economic system reacts instantly to the external driving im
pulse In terms of my earher 51xcategorxes the system 15his
toncal but static
(2) At the other extreme there ISthe purely endogenous theory
or the so called sell generatlng cycle Vanous theories emphamz
mg monetary factors 1nventor1es the acceleratlon pnncnple psy
chology etc fall under this heading The determmants of the
system cons1st of dynannc equatxons Involvmg dlfferent penods of
time (lags denvatlves etc) Which generate recurrent mot1ons
Once the cycle xsunder way boom nges Way to depression depres
snonto revwal rev1val to boom and so forth
Some economlsts thmk that 11leg1t1mateCircular reasomng IS
Involved m this View that It shows how the cycle once started
perpetuates 1tself but begs the questlon as to the ongma! ex1stence
of the cycle Even ll valld thls would be an object1on ol no par
thular consequence smce there are an mmty of exogenousand
chance factors whnchcould gwe rise to the :mtxal cycle Actually
as Wlll be shown below a cycle as not reqmred to start off the
process but only an mltxaldzsplacementfrom ethbnum however
small and noncychc character
ENDOGENOUS MODELS
Analytxcally the pure endogenous cycle 15 usually hkened to
the manon of a frxctlonless pendulum wh1ch satlses a sunple
Newtonlan second order differential equatlon Upon closer ex
ammatnon dlfcultles appear w1th thls notnon In the rst place
all dampemng must be ruled out or else the cycle wdl come to an
end Slmarly 1n most theorles ann dampenmg or exploswe be
hawor xs ruled out Now m a physmal system there are grand
conservatlon laws of nature whtch guarantee that the system
must fall on the thm lme between dampemng and am: dampemng
between stability and Instablllty But there IS nothing 111the
economtc world correspondmg to these laws and so xt would seem
nnmtely Improbable that the coe'icrentsand structural relattons
of the system be Just such as to lead to zero damperung
There ls st1ll another dlculty Even thhm an endogenous
theory the economlc system ISnot regarded as nsolated It ISsub
1ect to OUtSIdedxsturbances but these are not taken to be related
to the busmess cycle m an Important way If we consndera system
which taken by Itself IS on the border lme between stabllltY and
instability, it may be said to be stable in the second sense discussed
earlier. But if we now let random variations impinge on the sys
tem, it becomes unstable in the sense that the (expected) amplitude
of its swings increases with time, and its variance approaches
innity in the limit.33
I think it ill-advised, therefore, that Kalecki in an empirical
determination of the coefcients of an assumed mixed dierence
differential equation system should have imposed the conditions
that the motion have zero damping. One can sympathize with
his aim of deriving a system with constant amplitude of uctuation,
whatever one thinks of the empirical foundation for this hypothesis.
But after imposing a highly improbable constraint, he did not
achieve what he was looking for; namely, a system with constant
amplitude of uctuation.
Equally important, the purely endogenous analyses so far dis
cussed are unable to provide an explanation for the exact or ap
proximate amplitude of the cycle. Like the pendulum system they
are essentially linear in character, and every linear system can take
on any amplitude depending only upon the magnitude and sign of
the original displacements. Thus, a pendulum can be made to
swing in a small or large arc by means of a small or large initial
displacement. To explain the observed level of oscillation of the
cycle the economist with a linear endogenous theory must go back
to the value of the original prehistoric disturbance, and must ex
plain why succeeding shocks have not swelled the cycle.
There are two avenues of escape from the fundamental dii
culties encountered in the simple, linear, purely endogenous models.
The rst, which has been widely discussed in the technical litera
ture, involves dropping the assumption that the system possesses
zero dampening, and placing reliance upon external shocks to keep
uctuations from dying down. This solution involves dropping
the assumption of a purely endogenous system, but permits the
retention of linearity assumptions. It is discussed in detail in
later sections. It is to be stressed that the exogenous impulses
which keep the cycle alive need not themselves be even quasi
oscillatory in character.
On p. 343 below the sum of the A's will not converge nor will the sum of their
squares, where there is zero dampening. _
" M. Kalecki, A Macrodynarnic Theory of Business Cycles," Ecanamemca, III
(1935), 327-352.
The other alternatwe ISto drop the assumptxon of hnearzty even
though thrs entails cons1derablemathematical drfcultles For the
qualltatwe problem of stability, relatlons m the small were seen
to have Important lmphcatxons as necessary condttuons for stabthty
or Instablhty m the large In consequence 1t was useful and
permxssrbleto work thh hnear relations as a rst approumatton
even though that approxmntlon was admittedly not exact But
an the realm of quant1tat1ve dynamncs and economtc uctuahons
this approxunatlon must be dropped ll we are to get correct quan
txtatwe and Quahtatwe results
For hnear systems are lachg the quahtatxve rlchncss of
nonlmear systems The former are ezther damped or undamped
stable or unstable regardless of the sxze of lmtlal displacement
Nonlmear systemsfor therst lame mtroducc a theory winch accounts
for uctuatwns of a partzcular amplztude, mdepcndent of the mmal
dts-placement Ihus a nonlmear system may possess an unstable
stationary lexel so that when the cqulllbnum posmon rs dISplaced
m the sllghtest exer mereasmg swmgs are begun But Instead of
oscxllatmg to 1nn1ty, a partncular amplitude ISnally reached and
mamtamcd Thts penodtc motton may be stable m the.sensethat
any further dtsturbance wall result a motion hxch approaches
the gwen penodlc motton from above or below
As an example of such a nonlinear equatlon. consxder the
dlfferentzal equation
xp(1x)z+x=0 (53)
While zero IS a stat10nary eqmlxbrlum lewcl of the system an
evaluat1on of the coeicxents of the aboxe equation for that level
wnll show that the equnlnbrlum xs unstable For small departures
from equlhbnum the system behaves hke an exploswely oscxllatory
lmear system but nally the nonlmeanty asserts Itself and the
motlon settles down to a xed amphtude Geometncally, 111the
(x dx/dt) phase space the ethbnum pomt ISgwen by the ongm
out from whlch ow expandmg spirals However, each of these
approaches a closed curve representmg the periodic motion If WC
begm mltxally from a pomt outsnde the closed curve we have con
tractmg splrals wh1ch approach the closed curve from w1thout
The stablhty of the perxodxcmotnon 1n the small 15venable by
examining the solutmn of a Vanatnonal l1near dt'erentlal equation
of the second order with periodic coefcients, derived by evaluating
the coefcients of (53) along the periodic motion in question. This
equation of variation in turn is seen to be stable because its
characteristic exponents or multipliers all have negative real
parts.35 The uniqueness of the periodic solution is more difcult
to establish, but in the case cited it has been done.
Such nonlinear systems have received a certain amount of
attention in theoretical mechanics under the headings of self
exciting oscillations or relaxation oscillations. 35 Much still
remains to be done in this eld. In the eld of economics one
nonlinear system has received complete treatment. I refer to the
so-called cobweb theorem in which supply lags one period. In
general, this leads to a nonlinear dierence equation of the rst
order, which is so simple as to permit of a complete graphical
solution.
As we have seen in previous chapters, the linear cobweb case
permits of only three possibilities: for any amplitude the system is
either damped, anti-damped, or exactly in between. (Actually in
real life, both in an economic or physical system, a movement to
innity is unthinkable. Therefore, if a linear system is unstable,
its motion will grow until it no longer remains linear; its structure
will give," etc.) Dropping the assumption of linearity, we nd
that the system may possess special periodic motions other than
that of stationary equilibrium. Thus, with the equilibrium point
unstable, there will be a stable box approached by all near-by
motions, the size of the box determining the unique amplitude of
the cycle. There may, of course, be several such periodic motions,
each being alternately stable and unstable. In real life the last
one can in most cases be taken as stable.37
See G. D. Birkho, Dynamical Systems, chap. iii.
Equations of this type are sometimes known as Van der Pol equations after the
name of the man who has devoted a good deal of attention to them. However, their
history goes back into the 19th century at least. A partial bibliography n be found
in T. Von Krmn, The Engineer Grapples with Nonlinear Problems," Bulletin of the
American Mathematical Society, XLVI (1940), 615-683. See also Norman Levinson and
Oliver K. Smith, A General Equation for Relaxation Oscillations," Duke Mathematical
Journal, IX (1942), 382-403; B. Van der Pol, Relaxation Oscillations," Philosophical
Magazine, II (1926), 978992.
"There is a vast literature on the cobweb theorem. The most complete treatment
seems to be that of W. Leontief, Ventigerte Angebotsanpassung und Partielles Gleich
gewicht," Zeitschrift fr Nationaloknomie, Band V (1934).
34o F0 UNDATIONS OF ECONOMIC ANAL }SIS
Elsewhere m economic analys1s there are to be found suggestions
of nonhnmr dynamics Honcver the formal dlcultxcsof 5qu
tlon are so great th lt very much remains to be done [hlsls all
the more Important smce '1 careful analys1s of various hterary
theones wnllshow that some of the most sample of them depend m
m essentral 11y upon nonhnmr elements Among these ISa wnde
class of what I have termed blllludtable theories, whtch deduce
a tummg pomt of the cycle from such consuderatnonsas the fact
that the system llltSthe full employment celhng and bounces
back so to speak ' Another of thus type 15the Hawtrcyan nonon
of the minimum reserve ratios of banks against wlnch the system
rebounds Such notlons are less successful m explammg the lower
turning pomt sxnce there 15no (relevant) natural bottom to the
economic system 39
It Is not unexpected to nd that the Simplest emprncal notions
may lead to the most complicated mathematlcal problems Thxs
1s a fact to mspne hunnhty m both hterary and mathemahcal
Investlgators but should prove discouragmg to neither
Mumu YOGDINOUSENDOGENOUS TIIEORIES
It ISnot necessary to subsenbe cvcluswely to one of these two

polar types Perhaps most economxsts are eclectnc and prefer a
combmatton of both For example an economlst who beheved m
the reahty of waves of dnfferent length mnght plausnbly regard the
long Kondratteff n ave as bemg primanly ewgenous m character.
dependmg as 1t does upon wars gold discoveries, and great tech
nologxcalupheavals Its Impact upon the system might mvolvc
endogenous transmnt cychcal movements but their movements
mxght take place m a much shorter txme period so that compared
" See Ph Le Corbe ller Les Systemes Autoentretenus et les Osalhtuons dc Reina
tlon Economelnca I (1933) 328 332 J Tmbergen Slammal Testing of Business
Cycle Throne: (Geneva League of Nations 1939)
Professor A. H Hansen m hs Fuca! Polacy and Brumes: Cycles (New York
Norton 1941) chap x11 seems to regard the level at wh ch the average propenSIt)to
consume 15100 per cent as such a natural bottom ] thmk th 3 :s puttmg the matter
much too strongly even though I agree that below th:s level oerta n upward tendencaes
are brought Into play R F Harrod s The Trade Cycle(Oxford I936) plac oondef
able emphas s upon nonlinear factors m connecnon thh h s dynarmc determmants
to the length of the Kondratie cycle these might be regarded as
rapidly damped and be neglected.0
On the other hand, the extremely short uctuations, to the
extent that they exist, might be explained almost wholly in endog
enous terms with reference to the dynamics of inventories, the
acceleration principle, speculation, etc. These short cycles would
perhaps die down were it not for the random and systematic dis
turbances received from the longer movements in investment.
To illustrate the above remarks the reader may construct for
himself a model in which there is superimposed a regular periodic
movement of autonomous net investment upon a system in which
both the multiplier and the acceleration principle are operating.
The period of the autonomous movement should be longer than
the intrinsic period of the (damped) responding mechanism. Spe
cically, consider the equation
Y(t + 2) a(1 + l5?)1(t+ 1) + aBYU) = PG). (54)
where a is the marginal propensity to consume, B is the so-called
relation" of the acceleration principle, and P(t) is a periodic move
ment, not necessarily a pure sine wave."1
If the process has been going along for a long time, it will nally
approach a periodic motion of income in which autonomous invest
ment shows a lead over income, and in which the amplitude of the
nal motion is proportional to that of P(t), the exact factor of pro
portionality increasing with the nearness of the intrinsic period of
the left-hand side to the postulated periodicity of the right-hand
side, i.e., with the approach to resonance.
However, if we break in at an arbitrary point of time, there will
be in addition to this motion a damped transient whosequalitative
properties depend only on the endogenous response characteristic
On the whole, we should rather expect greater irregularity in amplitude and
periodicity in a cycle generated by eXOgenousfactors than in one endogenous in character.
This is in accord with the prevailing view that a diversity of factors is responsible for
the few recorded "long waves" of economic history, and with the notion that there is
much less predictability into the future of even the qualitativefeatures of such a move
ment. On all these matters the reader may be referred to the wellknownviews of
Schumpeter, Mitchell, Hansen, et al.
Cf. P. A. Samuelson, Interactions between the Multiplier Analysis and the
Principle of Acceleration," Review of Economic Statistics, XXI (1939), 7578. E. G.
Bennion has worked out a number of interesting arithmetical models in which the agree
ment between theory and the model sequences is complete.
of the multiplier-relation system. This can give shorter cycles,
brought into life so to speak by the shocks incident to the longer
v.ave."
MIXED Svsrsxis OF A LINEAR STOCHASTICTYPE
Thus far I have been considering mixed exogenous-endogenous
systems in which the exogenous forces are periodic in nature, or at
least quasi-periodic. There is, however, a stochastical theory de
signed to explain the existence of quasi-periodic cycles by means of
a damped system responding to random shocks The latter serve
to keep alive the uctuations of the system in spite of dampening.
But in domg so they tend to displace the phase of the given motion
so that ordinary periodogram analysis Will not, in a long series,
reveal a signicant" period in the vicinity of the intrinsrc fre
quencies of the responding system. This is intuitively obvious if
we think of ordinary periodogram analysis as Fourier analy515,and
the latter as equivalent to tting by least squares the best single
harmonic to the time series in question. The ordinate of the usual
periodogram is equal to that part of the total variance (in absolute
or percentage terms) of the time series which can be explained by
the best harmonic of that frequency Because of the constant dis
turbance of phase, in a long series no sine wave will give a good t.
It remains an open question whether other of the usual methods
of time series analysis (counting distances between peaks and
troughs, etc) will sufce to give back the known periods even in
articially constructed model sequences of the type of equation (I)
If we drop the assumption of linearity, then the response of the system I show
some differences The amplitude of income will not be Simply proportional to the
amplitude of the function Pa), nor Willthe nal solution be a Simple additive combine!
tion of periodic and transient components. Qualitatively, however, the result will be
a nal approach to a periodic motion With characteristic shorter waves in the transmon
period Unless kept alive, these Willthe out
1938) H. \Vold, A Study m the Analysis of Stationary Time Series (Uppsala, Sweden,
Cl. the reference given in footnote 43 See also the brilliant contribution of
Professor Ragnar Frisch to the Cassel volume, Propagation Problems and Impulse
Problems in Dynamic Economics, EconomzcEssays in Honor of Gustav Cassel (London
1933). pp 171205. While ordinary periodogram analySis will not do the Generallzed
Harmonic Analysis" of Norbert Wiener is designed preCisely for problems of this type
See the references to his 1930 Acta Mathematics article in H. T. Dams, The Analym of
Economic Time Series (Bloomington. Indiana Principia Press, 1942), and 3130the refer
case there to the 1935 article of Barrels. and the latter's suggestive concept Of the
"harmonic dial "
F UNDA 11!EN TALS OF D YNAMI CAL THEOR Y 343
above. It is clear, however, from the work of Slutsky 5and others
that the time series generated by such sequences resemble quali
tatively the usually encountered economic time series.
By means of the analysis of the previous chapter it is clear that
(except for terminal adjustments which become negligible in a long
series) the solution of the damped dynamic stochastic system of
the following type
L(Y) = Y(t) +a1Y(t 1) + +anY(t 72)= {z,}, (55)
where z is a random variable serially uncorrelated in time and
drawn from an unchanging universe whose rst two moments
(0, fr;-)exist. takes the following form
Y(t) = A0Z(t) +A1Z(t 1) + +AnZ(t -n) + (56)
As t grows the number of coefcients in this sequence becomes in
nite, but in such a way as to leave their sum and the sum of their
squares nite and equal respectively to (1/2": a,,
le). By the usual
0
central limit theories it is easy to show that (except for terminal
adjustments) the variance of Y(t) is given by
v, = (A02+ As + --+ ARM. (57)
or in the limit by
Ven: k}, (58)
where k is nite. The actual frequency distribution of Y(t) tends
to some limit with increasing t, with zero mean and variance equal
to the last given expressions. This is usually true even if the orig
inal frequency distribution of z is not at all normal (i.e., Gaussian).
The interested reader can satisfy himself that the superposition
of a random variable on to a periodic force leads to a forced motion
Eugen Slutsky, "The Summation of Random Causes as the Source of Cyclic
Processes," Ecanametn'ca, V (1937), 105-146.
The article by Trygve Haavelmo, The Probability Approach in Econometrics,"
Econometrica, vol. XII, Supplement (1944), deals with the problems ol' determining
empirically such stochastl relations. The article by H. B. Mann and A. Wald, On
the Statistical Treatment of Linear Stochastic Difference Equations," Ecanometrica,
XI (1943), 173220,shows that the conventional least squares treatment of autocorrela
tion is (asymptotically) a consistent" method of detemiining the a coefcients. It
would be outside the scope of the present work to go into these problems. A number
of articles in the Annals of Mathematical Statistics for 1942 deal with the sampling
distribution of the autocorrelation coefcient.
344 FOUNDATIONSor ECOAOMICANALISIS
Just hke that described above except that the mean of the asymp
totlcally normal distnbutxon uctuates according to the penodu:
functlon descrlbed 111the prenons sectlon He can also ork out
the lmphcatrons of havmg a stochastic variable 2; whrch IS not
serially 1ndependent
NON LINEAR Srocnasnc SYSTEMS

A more difcult problem xs that of developing for a non lmear
system the stochast1c theory wlnch corresponds to that dealt 1th
m the prevrous sectxon As far as the present wntcr ISan are this
15almost completely unexplored ground No more than a bird 5
eye vrew of the problem can be attempted here
We consrder a non lmear system of the form
10) *- IEYU '- 1) . YU n). 20)] = 0 (59)
\\.hose mmal conditions may be written tn the abbrevxated matrix
form
Yo = [Y(- t)] (60)
where z goes from zero to (n 1) If desu'ed the Y s and 2s may
be consrdered as column matrrces of many \anables As before
the Z refers to a random variable drawn from the same universe
over trme and mthout senal correlatron Formally the solutlon
may be wntten m the form
Y) = FEZ) Z " 1)! ! Z(0)| Y0] (61)
where the exact nature of F depends upon that off It has been
shown m Mathematical Appendn: B, Section 8 that mth an ap
propnate redemtron of the \armbles Y and Z ue may consrder
equations (59) to be of the rst order, 1e to depend only upon the
value of Yone perlod ago The new Z Willnow has e zeros m some
of Its components, but as before there wrll be no serlal correlatlon
between successwe Z 5
Under certam restnctlons upon the partial denvatxves off such
as correSpond to the economic reality of any relaxant damped eco
nomncsystem It should be possnbleto enuncnate hmnt theorems not
unhke those holdmg for lmear systems The fact that so many
ell known non lmear statrstrcs go to normahty m the hm1t as the
sample Suze becomes large suggests that many of these hmtt
theorems would actually be of Gaussian form." It would be a
task of some delicacy to establish conditions under which this must
certainly be so. However, it is not to be thought that in most cases
the limiting distributions must be of Gaussian form. In a moment
I shall specify a non-linear system which approaches a limiting
distribution not of Gaussian form; in fact, an example will be given
of one which remains bounded but approaches no stationary (prob
ability) state, but oscillates indenitely in a simple periodic
fashion.
The key to a successful analysis of the difcult non-linear case
lies in shifting our attack from the study of any one particular
motion as it is bombarded by a particular set of random shocks to
the analysis of the probability states corresponding to all possible
shock patterns weighted in accordance with their likelihood.
These contrasting approaches are to a certain extent similar to the
contrast between the motion of a single molecule, the totality of
motions of an ensemble of molecules as in statistical mechanics, and
the kinetic theory of gases which describes macroscopic states of a
system. However, the analogy is not perfectly complete and
should not be pressed too far.
First, let us assume that the initial conditions at time t are
known. Then from our knowledge of the probability universe of
Z. we may write down immediately the conditional probability
description of Y., given Y,, namely,
P(Y.+1, Y!) = H(Y,+1, Y.), (62)
where the exact form of H can be easily specied as soon as f and
the probability distribution of Z. are given.
Now if we assume that the exact probability distribution of Y.
is known, and equal to P(Y;), it is possible to write down the
distribution of Y.. in the form
P+1(Y+1) = : H( Y+1,Y)P1(Y)dY1. (63)

It is to be noted that even in the non-linear case the probabilities
at different instants of time are related by a recursive linear
functional.
Unlike most statistics computed from a sample, the F operations on the 2's are
denitely not symmetric functions. .
'I.
346 FOUNDATIONS OF ECONOMIC ANALSIS
For concreteness we may show what form the H takes m the
Simplest case of a one vanable rst order system of the form
y "= a)" + Z: (64)
wrth probabzhty denszty forZ gn en by R(Z) Then correspondzng
to equation (63) we have the relation
Puma) = f_ RUM cyaneody. (65)

To see whether or not P. approaches a hmxtmg dnstnbutxonxt
:s natural to put the same P m both sxdcsof the above equatton
and solve the resultmg equatton for the form of the unknonn dns
tnbutxon This 15an Integral equatlon of the Fredholm type but
thh Innite I1m1ts For the specral type of R gwen aboxe and
for a less than umty, the emstence of a hunt ls assured
For absolute a greater than unity we know that the variancem
creases Without hmlt. and that there cannot be a llmztmg form
other than the tnvnal zero solution We may state tlus m another
manner If he put a parameter A un front of the Integral m (63)
and ask for a function P Wthh satises both Sldes, such ml] exrst
only If A ns a characteristic value or eugen-value of the kernel II
For undampcd hnear systems A = 1 ml! not be a charactenstlc
\aIue It would be a task of some consnderable mathematical
dlfculty to Indrcate Just when the xntegral equatton ansmg from
a non hnear system had an exgenvalue equal to unity and to deme
the corresponding elgen function P(t)
The problem \\ ould be 51mphed 1f c could assume that our
non lmearf gn es bounded talucs, wlntexer the value of prenons
Y s Thls [5ob\ nouslyposmble only for non hncar systems, earlier
discusswn of the physncal [mutations of full employment and zero
mcome suggest that xt lS often reahstrc to assume burners \thtch
gammes ns fact Kathns assampt. '2 the absohete : 1'99 f
1must be less than some number M, and \\c hue M and +M
as the hnnts m the Integrals Instead of the mmte hnuts This
remoxes the smguhnty m the Fredholm equation It 15to be
hoped that thls contmuous but bounded case mll be fully anal) zed
by mathematxelans and economxsts
I shall conne myself to the snmpler case n here If 15a dlscrcte
rather than a contnnuous \arlable By making the classxcatlon
ne enough, we can reach any empirical degree of approximation
to reality so there is no essential loss of generality involved. By
replacing the above integrals by Stieltjes integrals, both cases can
be handled simultaneously. However, in the simpler case under
discussion, the integrals can be described by sums. But in this
case Y takes on only integral values, and the probability functions
for Y and Z are at each time denumerable sequences of numbers.
Corresponding to the kernel H we now have a matrix H whose
properties depend upon the f function and the probability se
quence R(Z).
However, to make the problem manageable there are other
difculties to be taken care of. Even in the simple linear case,
with Z taking on integral values and the initial Ys taking on
integral values, the Ys at a later stage will not be conned to
integral values unless the coefcients of the integral equation are
themselves integers. Let us satisfy this requirement in our simple
linear example of equation (64) by setting a equal to unity. Our
H matrix is then seen to consist of an innite number of rows and
columns, each column consisting of the R sequence of probabilities
referring to the different Z values, with R(O) centered in the
diagonal of the matrix as follows:
(66)
It is not an easy matter to work rigorously with innite matrices

and their latent roots and vectors. In this case we can be sure
that there is a latent root of unity by virtue of the property that
probabilities must add up to unity, but we cannot attach unam
biguous meaning to this, nor solve for the corresponding latent
vector. .
It might be thought that the difculty is of our own making
in that the R sequence has been considered to be innite. How
ever, the assumption of a nite number of terms in R only Intro
duces zeros in each column after one reaches a certain distance
from the diagonal. The matrix must still be considered to be
348 FOUNDATIONS OF ECOVOMIC ANALISIS
mmte 1n sue and nf we walt long enough Y can take on any
mtegral value however large
There ls the further more senous difculty that a damped sys
tem w1th leadmg coefcnentequal to umty cannot have coecnents
whrch are all Integers Clearly an equation whose roots are all less
than umty m absolute value WI" have a product less than umt}
and hence a coefcnent which 15not an Integer
Therefore even the discrete case I shall make the non lmear
assumption that the differenceequation whichdenes the dynamacal
path of the system 15such as to yleld bounded values of Y. It IS
convement to assume that the Y s m addltton to being discrete tal.e
on only a mte number of values thts bounded mten al 1e e
round off to some degree of accuracy Because we make our
class1catxonsas ne as we Wish there 15no senous loss of generallty
Involved In this case the probablhtles at time t of the different
values of Y n m number can be represented by the equation
P+1 = HP: (67)
where the P s are column matrlces of n elements and the H ISa
square 1: by n matrlx whose propertles depend upon those of the
non lmear system and on the probabnhty dxstnbutxon of Z It
wnllbe noted that the sums of the columns ol H are m every case
umty Thls follows from the con51deratxon that nf Y 15certamly m
posmon : at a given tune (so that the P vector has zeroseverywhere
except for a umty 1n the zth element) then It must certainly be
somewhereone period later But Its probabxhty of bemg m each
of the n posmons one period later ls under these condltlons nothmg
but the 1th column whose sum must therefore add up to umty
Note also that all of the elements of H must be posxtne m vu'tue
of their mterpretatxon as conditional probablhtxes
Because the columns add up to umty we place a minus one In
each dragonal and add each row to the rst row The result 15a
smgular matnx W1thzeros m the rst row ThlS shows that umty
ISa latent root Slmllarly the fact that all elements are pOSItlve
and add up to umty guarantees that there are no latent roots which
exceed umty m absolute value For 1fthere were we could select
lmtlal condmons for whlch a given element P would grow ex
ponentnally w1thout llmlt Thls contradlcts the assumptnon that
(probablhty) element of P can exceed umty
FUNDAMENTALS OF DYNAMIC/IL THEORY 349
In most cases there will be one root of unity and all other roots
less than unity in absolute value. Thus, whatever the original
probability distribution of Y, it will gradually approach a station
ary probability state given by the latent vector of H corresponding
to the unit latent root. It will satisfy the equations
HP = P, (68)
where the P is normalized so that its elements (and not their
squares) add up to unity. The exact values of the Ps can be de
rived by solving (7: 1) linear equations. It will be noted that
the solution secured will have nothing to do with the limiting
Gaussian form.
In special cases the unit latent root may not be simple so that
no unique stationary probability distribiition is reached. In still
other cases there may be another latent root whose absolute value
is equal to unity, and which is either complex or equal to minus one.
In either case for most initial probability states there will be no
approach to a stationary probability state; instead there will be
periodic oscillation. A simple example is provided by a 2 by 2 H
matrix of the form
0 1
[1 0 ]. (69)
If we start out with a probability vector (a, b), it gives rise, because
of the negative latent root, to the oscillatory sequence (b, a), (a, b),
(b, a), etc. , without ever approaching a limit.
See W. Feller, An Introduction to Mathematical Probability and Ils Applications (New
York: Wiley, 1950) for discussion of Marko processes.
CHAPTER XII
CONCLUSION
ECONomcs 18a growmg subject m wlnch very much rs left to be
done It ISonly appropnate, therefore, m brmgmg thrs u.ork to a
close, to mdncate a few of the 1mportant unsolved problems crymg
out for further 1nvest1gat1on
In the rst and second chapters I outhned the general problem
of comparatwe statzcs how from a knowledge of the qualitative
and quantltatnve propertres of our eQurhbnum condltlons we can
hope to deduce meanmgful theorems concermng the dlrectron and
magmtude of changes m our variables when certam data change
In chapter m It was shown that m a large class of cases the econo
mrst derives denlte theorems by means of the hypothe51sthat the
equlhbnum posntlonrepresents a manmum or mlmmum posmon
The mequahtzes assocnated With the demtlon of an extremum
posxtronwere seen to be the source of frultful theorems m compara
twe statxcs
Chapter w represented an apphcatson of thxs analysrs to the
cost and productlon theory of the rm, Just as chapter v gave a
treatment of constramed maxrma as requrred by the theory of con
sumers behavror Specxal aspects of the latter subject recewed
treatment m chapters vxand vu And nally the study of statlc
maxnma and mrmma was brought to completion m the analyms of
welfare economies given m Chapter v
In the rst chapter of Part II It was shown that s1mplyfrom
the standpomt of fruxtful comparatrve statics dynamzc analyszs IS
useful and necessary Indeed, the correspondence prmczple, emm
cxatmg the relatlonshrp between the stablhty condltlons of dy
namtcs and the eva'luatlon o't dtsp'lacements m comparatwe s'taucs
prOVldeSthe second great weapon the arsenal of the economlsl
mterested m dernrng demte, meanmgful theorems
In chapter 1 dynarmcal systems are studned for therr own sake
particularly un thelr stablhty aspects whlle m chapter x1 I have
discussed Various fundamentals of dynamical analysns mcludmg
1 problems arxsmg out of the study of busmess cycles
350
CONCLUSION 351
Broadly speaking, the development of analytical economicshas
proceeded in a natural evolutionary order. First, in \Valras we
have the nal culmination of the notion of determinatenessof equi
librium on the statical level. This has received further elaboration
in the hands of Pareto and others.
However, Pareto took a second, further step. He laid the basis
for a theory of comparative statics by showing how a change in a
datum would displace the position of equilibrium. Even earlier
Cournot had pioneered in this innitesimal analysis, although
with reference to a narrower set of problems.
While Pareto laid the basis for comparativestatics, his own work
was not rich in denite theorems in this subject, precisely because
he rarely concerned himself with the secondary inequalities relevant
to maximum positions. On the few occasions when he did do so,
he came to grief because of mathematical errors in their statement.
It was left for W. E. Johnson, Slutsky, Hicks and Allen, Georgescu
Roegen, Hotelling, and other modern writers to begin to make
progress along this third line.
However, only a part of economic theory is concerned with the
maximizing action within an economic unit. Where the inter
actions between individuals are concerned, the scope of fruitful
comparative statics may be greatly extended by a fourth advance,
the apprehension of the correspondenceprinciple, whereby the com
parative statical behavior of a system is seen to be closely related
to its dynamical stability properties.
A natural fth step to take after we have investigated the
response of a system to change in given parameters is to investigate
its behavior as a result of the passage of time. Thus, we cultivate
dynamics for its own sake, especially with respect to the qualitative
properties of the respective motions.
The usefulness of any theoretical structure lies in the light
which it throws upon the way economic variables will change when
there is a change in some datum or parameter. This commonplace
holds as well in the realm of dynamics as in statics. It is a logical
next step, therefore, to begin to create a theory of comparative
dynamics. This will include the theory of comparative statics as
a special case, and indeed all of the earlier ve subjects, but it Will
cover a much richer terrain.
The central notion of comparative dynamics is simple enough.
We change somethmg (Just what need not concern us at the mo
ment) and we mvcstlgate the effect of thus change on the whole
motion or behawor over tune of the econormc system under
vestlgatlon It wxllbe seen that comparatne statlcs Involves the
specral case where a permanent ' change 18made, and only the
effects upon nal levels of statlomry equtllbrlum are m questmn
In comparatwe dynamzcs we conszder a much broader class of
changes (a) We may make a change m mmal condztzons By
demtlon thzs alters the Immediate behauor of the system m a
known way By the assumptlon of contmulty we may mfer that
the posmon of the system for some region adjacent to the mtttal
condntlons ISalso altered m the same drrectxon For Intermediate
lapses of txme a separate Investxgatlon IS necessary to determme
what happens to the system However for a stable system It 15
clear by Virtue of the denxtnonof stablhty that for sucntly
long txme penods there wxll be no nal alter-anon m the behavnor
of the system
(b) We may make a change m someforce acting on the system
Thus we may cause autonomous Investment to vary Actually,
there are a vanety of cases which must be consndered The change
1n force may be permanent, It may be Intermlttent, 1t may be
transnent or Instantaneous In this very last case the analysrsmay
be subsumed under the headmg of a shift m nmtmlcondltlons In
the case of stable systems the response to a permanent alteratnon
glves us a descnptron of the actual path followed by a system 111
gomg from one comparatlve statlc level" to another
For hnear dynamrc systems but only {orthese, the most general
of the above vanat1ons can be thought of as made up of the cumu
lated effect of umt Impulses, or of changes m the mstantaneous
lmtxal condmons of the system Thrs follows from the base
superposztzon theorem which underlies much of apphed mathe
matrcal analysrs
(c) Fmally, there may be a change m some internal parameter
of the system We may ask, for example, what the effect of a
cha age m the margmal propens1ty to consume or 111the relatxon
may have on the behavxor of a system Agam, the change ln
question may be permanent, varying tranSIent, etc
The nch variety of forms whlch the change m data may take
matched by the numerous ways m wh1ch we can choose to de
CONCLUSION 353
scribe the resulting eects on the behavior of the system. Ex
cept in the most simple cases we shall nd it necessary to summarize
in various ways the information contained in the resulting changes
in the system at each and every instant of time.
From a short run point of view interest will center on the im
mediate reaction on the system. The answer to this question can
often be secured by a method which is formally like that of com
parative statics with the signicant difference that some of the
variables treated are really dynamic in character.
This may be illustrated by the important case in the Keynesian
analysis where investment is taken to be a variable whose value is
to be determined by a system of relations like those discussed at
the end of chapter ix. For the usual Keynesian theory, in the
short run at least, the stock of capital is taken to be constant.
By solving such a system we can nally reduce our relations
down to a single equation between the amount of investment, I,
and the stock of capital, K, and the value of some parameter, a.
This parameter might well be the amount of thriftiness in the
system. Now if we are interested in the effect upon the stock of
capital in the immediate future of a change in this parameter, the
answer can be given by treating investment as if it were an ordinary
statical variable and solving, as in chapter ii, our equilibrium equa
tions to determine the direction of change of investment with
respect to the parameter, a.
However, to answer our question in comparative dynamics we
must introduce the fact that investment, which we have previously
been treating as an ordinary statical variable, is really the rate of
change of the stock of capital, or equal to dK/dt. If the compara
tive statical analysis tells us that investment is reduced by an in
crease in thriftiness, we can be sure that in some sufciently short
run the amount of capital in existence will be less than it would
otherwise have been. For if two curves start out from the same
point with diering rates of increase, we can be sure that the one
with the greatest rate of increase will exceed the other for some
small region at least.
As it happens, a similar statement can be given in this case to
the wider problem as to what happens to capital in the long run as
a result of a change in thriftiness. Thus, if instead of simply asking
what level of consumption maximizes current investment, we
354 FOUNDATIOAS OF ECONOMIC ANALISIS
mden Professor Lange s questlon 1and seek the leels of consump
tnon Ieadmg to the most capital at each mstant of tzme ue shall nd
that capital formatnon m a mn of any length ISonly maurmzed If
at each Instant the Lange crlterxa are met
Hon ever when the problems of comparative dynamics are
posed m general and reahsttc form 1t turns out that the aboxe
1dent1tyresults almost by chance Thus con51dera system of tuo
or 12dynamlcal equations
% = g(x1 x,. a) (z _ 1 n) (1)

Suppose furthermore that an Increase m a alwa}s Increasesde/dt
or that ,g..1lS alu ays posmve Does It follow that an Increase m 0:
always results m a hxgher xl at each subsequent Instant of tune?
The answer IS no For some suicxently small Instant of txme
starting out from gwen 1mt1alcondntlons thus must of course be
true but xt need not contmue to be true In the one vanable case
to whxchI reduced the Keynes Lange system the stronger theorem
IS true For m thts case we may solve etphmtly our dtfferenual
equatlon the form
! du
tfJ:m=0 (2)
and by parttal dlfferentlatlon of the above Impllcntrelatlon between
t x1 and a It xseasy to show that the change unx1w1th respect to a
t being \ed must be of the same Slgn as the demtely known
coefcxent g..l The reader can u ork out the geometrtcal reason
for thxs
Not only 15the direct solutlon of the system 1mposs1ble the
multx vanable case but the corresponding theorem ls demtely
untrue Economtcally th1sis not hard to Visuahze If the param
ete: & has a 'p'itmoumedeettupon the growth 81a mend van
able after a sufcnently long penod of tlme thls mdu'ect Inuence
may outwelgh the dlrect favorable Inuence on the rst variable
The reader may w1shto \\ orL out an even more comphcated model
than the above m whlch a enters as a dynam1c parameter Thus
xt nsnot dxfcult to construct a model ln whlch dec1t nancmg has
'0 Lange The Rate of Interest and the Opt mum Propens ty to Consume
ca \ (1938) 12 32
CONCLUSION 355
a favorable effect upon the growth of capital in the short run, but
the accumulated bad effects of the growing debt are adverse to
capital growth. I do not wish to pass judgment upon the reality
of the above assumptions, but simply wish to point out the possible
occurrence in economic systems of the common medical phenomena
whereby short term remedies may have long term deleterious
effects.
Of course, if we conne our attention to the behavior of long
period stationary equilibrium positions, the methods of compara
tive statics come into their own again. All time derivatives, differ
ences, etc., are set equal to zero and the resulting system solved
like any staticaI system of chapter ii.
Finally, the economist will often be interested in the effect upon
some characteristic of the motion of the system. How will a
change in the relation affect the average level of a uctuating
trendless system, its periodicity, its dampening, and its amplitude?
Nor need we conne ourselves to simple stable systems. Thus, it
is of the greatest interest to know how a change in age-specic
mortality will affect the net-reproductive-rate; or how the period
of delay in the expenditure of income will affect the rate of ination
under a given dened inationary gap.
The further development of analytical economicsalong the lines
of comparative dynamics must rest with the future. It is to be
hoped that it will aid in the attack upon diverse problemsfrom
the trivial behavior of a single small commodity, to the uctuations
of important components of the business cycle, and even to the
majestic problems of economic development.
MATHEMATICAL APPENDIX A
MANY MATHEMATICAL
THEOREMSof importance to the economist
are not to be found collected in any one convenient place. W'ith
out striving for rigor, I have gathered together some of those most
relevant to the matters under discussion in the present work.
I. MAXIMUM CONDITIONS FOR A SINGLE-VARIABLE FUNCTION
Let
z = f(t)
be a dened function with a continuous second derivative every
where on an interval
a HA t b.
2 will enjoy a relative maximum at the point t, providing
f(t) EN), a < t < b (1)
for neighboring values Oft.
It is necessary in order that this condition be satised that
f(t) = 0, (2)
and
f(t) 0- (3)
The rst Ofthese conditions is derived in the following manner.
By the Theorem of the Mean,
f(t) f(t) = (t t)f'[t + 0(t t)]. 0 < 0< 1
Suppose
f(t) > 0.
Then from our continuity assumptions there exists an interval
[t t] < m
within which the following inequality holds everywhere
f (t) > 0.
Hence, for all such neighboring values Oft larger than t, we have
f(t) - N) > 0,
357
which contradicts our hypothesis. Similarly,
j'm) < 0
would lead to a contradiction. Therefore,
f (t) = 0.
The second necessary condition is derived in a similar manner.
From Taylor's expansion with a remainder we have
f(t) m") = (: t)f'(t)+ _;ff'w + 9%:m.

0 < 9 < 1
But
m") = o.
and so this becomes
fu) N) = f2f" + m - m. 0 < 0' < 1

Suppose
f '' (t) > 0.
Then by assumed continuity
f"() > 0
on an interval
[t -t! < m'.
In consequence,
f(t) N) > 0
on this interval. But this is a contradiction. Therefore,
f(t) 0.
Actually, it can be shown under conditions of proper continuity
that the rst non-vanishing derivative must be even and negative
{bra relative mam'mum.
I dene a regular maximum as one for which the equality sign
in equation (3) is excluded. It is necessary and sufcient in order
that z enjoy a regular maximum at t that
N") = 0.
and
f"(t) < 0. (4)
MAXIMA AND QUADRATIC FORMS 359
Similarly, necessary conditions for a relative minimum under
our assumed conditions are
f(t) = 0.
and
f(t) O. (5)
Necessary and sufcient conditions for a regular minimum are
f(t) = 0.
and
f(t) > 0. (6)
II. MAXIMUM CONDITIONS FOR A FUNCTION OF
MANY VARIABLES
Let
Z =f(x11 ' "1xn)
be a dened function with continuous second order partial deriva
tives of all kinds in an open region S. Then a point (X) in S
affords a relative maximum to 2 provided that
f(x1,'-,xn)gf(x10, "',x,.)
for sufciently close values of (X).
It is necessary in order for this to be true that
af<x10 . . . , an) _ 0
8x1
af(x1v v x110) =0
6x2
af(x107 ' ' ': xno) : 0

acn
and
> a2f ai'f af ___ f
0 : xlxl hlz + 2 xlxz hlhz + 2 xlxahlha + + 2 xlxnhh
__f ___.agf ... __a'2f
+ xzxg 122 + 2 axgax3 hzh + + 2 xzxn 2h
+ .............................
+................
Off 2
360 FOUNDATIONS OF ECOAOMIC ANALISIS
here (111.11,, , h)are arbztrary numbers, and the partial de
m'atn es are exaluated at the pomt (X)
If we adopt the followmg notatton
f(x:. . x..)
flo : fn(1loo . x130): __Tx? : (1 = I. p)
and
o _ f(x10, v xno)
fu _ xx, :
then our necessary condmons can be wntten symbolically
f. = 0. (z = I. .n) (7)
and
}; lefuthjz, ; O (8)
It ISnecessary and sulcnentfor a regular relative maximum that
fo : a (t : lt : )
and
);, 2f,,h,h,
1 <0 (9)
for not all h's equal to zero ln addltlon to the vamshmg of all
rst partial denvatlves a regular relatwe mammum requires that
the quadratlc form whose coefcxentsare the second part1al denva
tlves be negatwe denxte Thrs last condltron 15the generahzatlon
to many vanables of the Singlevanable condmon that the second
denvatxve be negative for a regular relatwe Simplemax1mum Its
meamng mll be amplred m a later sectron
The proof of these necessary condxtlons can be easrly demon
strated by the use of a devxce Which enables us to convert our
problem mto that of a single vanable one upon whaeh we can
bnng to bear the results of the prevnous sect1on
Let
x. = xg + u, (z = l, , n)
here the h's take on arbitrary values Clearly for a relatne
maxtmum xxemust hue
f(x10 + Il; xzo + th! , xno +1,13) f(xlov vxno)
MAXIM/l AND QUADRATIC FORMS 361
For given values of the hs, not all equal to zero,f can be regarded
as a function of the single variable t, and the above condition can
be written
f(t) f(0)
From the previous section we know that this requires that
f(0) = 0,
and
f(O) ; 0.
But, by performing the indicated differentiation, we nd
f(0) = fiohi = 0. (10)

and
fn) = ; fijohihj % 0.
In order that (10) be satised for arbitrary values of the hs, each
partial derivative must vanish, or
fio=0. (z'=1,--',n)
For a regular relative maximum, we get by this method
fe=0. ('i=1,'',n)
and
%, fijohihj < 0, not all h's = 0,

1 1
these conditions being both necessary and sufcient.

Similarly, a regular relative minimum requires that
f.-=0. (i=1.",n)
and
i if.,-h.-h,-> 0, not all hs = o. (11)

l 1
The appropriate quadratic form must be in this last case positive

denite. ,
III CONDITIONS FOR A CONSTRAINED MAXIMUM
As before, let
Z = f(xl | x)
be a dened function mth continuous second partlal denvatwes in
an open regxon S In addxtxon, we Impose subsxdlary condltxons
on the admlss1blevalues of our vanables In the form of the lollowmg
m xmpllcltfunctions of the n vat-tables,dened m S mth contmuous
second part1al derivatives,
G'(x., , x,.)
Gz(xl| cn) :
6"(361, , xn) : 0
Under what condltlons wnlla pomt (X) m S afford a relatwe
maxzmum to z, and satlsfy the subsndlary cond1tlons? Such a pomt
ISdened as follows
f(xlp , xn) : f(xloo ! x00)!
where
G"(x1, , x,.) = G"(x;, . x,?) = 0, (k = 1, ,m)
and (X) sucxently close to (X )
The way to solve thls problem whnchwould suggest itself most
naturally would be to use our m subsxdxarycondxttons to ehmlnate
m of our variables by expressmg them terms of the remammg
(n m) vanables We assume that the rank of the matnx
x. 6x; 6x.
& 92 66
83: x a ,.
x , or [G] (k=l, ,m)
(!=I, ,")
@ af: 86"
6x1 6x2 ax
ISof order m so that at least one m2determinant does not vanish
This insures us that we can in fact uniquely solve our subsidiary
conditions for m variables in terms of the remaining (n -m).
Since the numbering of our variables is arbitrary, we may write
this without loss of generality as
xi = f(x+1, xiii-{~21 ' ', xn)- (1: = 1, ' ', m)
By substitution in our original expression, 2 can now be expressed

in terms of the variables (xm+1.xm+2,- -, x,.) with no conditions of
constraint, and the analysis of the previous section can be applied.
However, there is a great loss of symmetry in such a procedure
since not all our variables are treated alike. Fortunately, by the
use of an artice which can be rigorously justied, it is possible to
derive a more symmetrical set of conditions.
Consider the Lagrangean-multiplier expression
H = H(x1v"'ixn) =f(x1) "':xn)
+ 1G1(x1,' ' ')xn) + ' ' + )me(x1y ' "a xn):
where the Asare constants, not all zero, whose values will be de
termined later. Obviously for values of (X) satisfying the sub
sidiary conditions indicated above, all terms after the rst must
vanish, making our new expression equivalent to z. This suggests
that for a maximum at (X)
H 6 aGl" G'
=H0=+h+ +m7:
=0,
(1:=11'rn) (12)
subject to
G"(x1.--,xn)
=0- (k=1"'m)
It is unnecessary for our purposes to indicate explicitly the rigorous
proof of this statement.1
The last two sets together constitute (n + m) equations
in the (n + m) unknowns (xl, - - ,x> - - -, )). Under the
proper restrictions the values of all the xs and Xsare uniquely
determined.
See Carathodory, Variationsrechnung und Partielle Di'erentlgleichungen Erster
Ordnung (Leipzig und Berlin, 1935), pp. 164189. The conditions of (12) are zdenucal
to those of (13) which follow. Carathodory shows that if some minor of (13) greater
than 77:were non-vanishing, then one couldby the implicit function theoremsolve
for some :: in terms of f, where f is both greater and less than the extremum value, I.
This is a contradiction.
364 FOUNDATIONS or ECONOMICANAMSIS
Flumme rst necessary conditions un be expressed mdc
pendently of the As Consxder the matnc
'_f @ icf 59.92
x; 6x1 xz 6x1
a; 361 aG aa"
QZ 53 MM h=L n)
' or [ext 33:] (13:1: ,m)
@; aal an aan
keT a: ex,. ax..J
The rank of thlS matm: must be equal to m 1e , every (m + 1)

determinant must vanish Thus
_QL 66*
x. x. = :d) GS
and so forth for all possnble combmatlons of (n) varlables taken
(m + 1) at a tlme These wnll yield (n -m) independent condx
tions on the partial denvatwes off and the G s
For concreteness consxder the case of a many variable function
to be mammlzed as of one subs1dnary condmon of restraint Ac
cordmg to the condmons mdlcated above there wxllresult at an
extremum tangency between the locus of restramt and a contour
locus of z (11 l) condmons famlhar m many branches of eco
nomlc theory
In addmon to the rst necessary condmons upon rst partial
denvatxves there are as m the prewous sectlons certam added
secondary condmons on the second partnl denvatues The fol
lowmg must hold
iinm
l 1
_ n n afo 626 86 <

" >,:?(ax,+axat + +""a_:,ax)h"-O (14)
n k.
z=o ( .)
M'AXIMA AND QUADRATIC FORMS 365
In summary, it is necessary and sufcient for a regular relative
constrained maximum that
Hi0=01 (1;=1,"',1l)
and
i flijollihj< 0
HM: 1
for
n. k
2 (??]!
- =0 not all hs = 0.
The meaning of these secondary conditions requiring the de

niteness of certain quadratic forms under linear restrictions will be
dealt with later.
IV. QUADRATIC FORMS
A real (homogeneous) quadratic form in n variables (/21,- - -, k,.)
is dened as follows:
QUH, -, kn) = 011/112+ Gizhihe + ' ' + alnhlhn +
a2lh2kl + (122h22+ + aZnhZhn +
anlhnhl + an'zhnh'2+ ' . + annhn2 !

where the as are real constants. A quadratic form in the single
variable aswould take the form
Q(x) = 0x2;
in two variables (x, y) it would be written
ax2+ bey + cf.
Since the order of multiplication is indifferent, it is clear that
any quadratic form can be made symmetrical, i.e.,
au = un: a = a'- (17)
For if not already symmetrical, we may rewrite it without affecting
its value as
" 'l a a.
_:____i
??? + Jih -h,
whnch ns symmetncal Unless otherwlse specned wc shall deal
only w1th symmetncal quadratic forms
A posmve demte quadratic form ISonc which Is always posmve
except when all the variables equal zero (at whxch pomt all quad
1'3th forms vamsh) A negatzve demte quadratic form :s one
which 15always negative except at the ongm A demte form xs
posxtwe demte or negative demte A semz-demte form xs one
which never changes sngnbut may be zero at a set of pomts other
than the ongm Thus a posztwe sem: demte form could more
appropriately be called a non negatwc form All forms whxch are
neither demte nor serm demtc are called mdemte
A quadrattc form m n vanables xs smgular 1f the rank of xts
matrix ?
a = [O.,]
18less than n. : e 1f the determmant
laul : 0
It may be observed that all semt-demteforms are smgular
Forms m less than three vanables

In the case of a smgle vanable, 1t 15 easy to determine the
demteness of
Q(x) = ax
The form 1s (l) posutxvedemte, (2) semi demtc and smgular, or
(3) negative demte depending upon whether
a 2z 0
It ls more difcult to ascertam whether a form m two vanables
ISposxtlve demte If
Q = ax2 + 2bxy + cry
18to be always posmve except at the orngm (0, O), 1t must clearly
be posxtlve for (1, 0) But
Q(l. 0) = a. (18)
and so a necessary condmon IS
a > 0 (19)
A matra)::: of rank r xrall Its determmants of order (r + l) vanish All detcnm
mat! of a sul! lugher order mll be found also to vanish
Similarly, it can be proved that
c > 0. (20)
These conditions are necessary but not sufcient. To derive
still another necessary condition and determine complete sufcient
conditions, let us consider the form as a quadratic equation in x
with parameter y,
f(x) = ax2 + (263026+ (car) = 0.
If the form is positive denite, x must not vanish for any non
vanishing y. Solving by the familiar rule for extracting the roots
of a quadratic, we have
_ Za
_ _ 2by :1: \i4b2y2-4acy2 : [- b :}: @
Vb? 06])!
But this must yield no real value of x or the form would not be
positive denite. Hence, the expression under the radical must be
negative, and we have as an additional necessary condition
ac b2 > 0. (21)
However, our three necessary conditions (19), (20), and (21) are
not all independent. The last of them in conjunction with either
of the rst two implies all three. The sufciency of these condi
tions is implied by the fact that the form is positive for the values
(1, O), zero only for (0, 0), and never changes sign; consequently,
it is positive denite.
These conditions may be given a more convenient form. The
matrix of the form in two variables is obviously
a b
[b cl
If we dene a principal minor of a determinant as a subdeterminant
formed by crossing out the same rows and columns so that the
diagonal elements of the minor are contained in the diagonal ele
ments of the original matrices, then our necessary and sufcient
conditions are implied in the statement: all principal minors must
be positive. These are three in number, but not being all inde
pendent, a complete set may be specied by making successive
principal minors positive starting with the upper left-hand corner
and adding successively corresponding rows and columns. (In this
case there is only one row and one column to add.)
{or the form to be neg'xtwc denite ll. could be shown along
sumzlarlmes that
a < 0
ab
bc > o. (22)
or that the rst prmcxpal mmor ISnegatwc, and the successwe ones
alternate m Sign those of odd order being negative, and those of
even order bemg posmve Thts ISbecause [a ,] nsnegative demte
only 11'[-a ,] ISposnttve demte
Farms m any number of tenable:
The specnalmethods applicable to forms of few variables cannot
be apphed m the general case We must therefore seek more
general methods A non smgular transformation of varrables of
the type
h...b.,p,. (1:1 ,n)
or usmg matrm notation,
h = bp, p = brh,
leaves the quadratxc form posrtwe demte If ongmally posxtne
demte Tlns ns made plausuble from the recognition that the
ongm goes mto the Ongm, and that every pomt m the h space xs
umquely related to every pomt 1n the 13space Consequently, Q
after the transformation takes on only the \alues tal-.cnon before
00!) = (2(1)?) = @(P)
In partzcuhr let the transformation 6 be such as to take Q mto
a sum of squares of the form
Q(bp)= aw + dm; + + (hp,? (23)
lt ISwell known that an mmty of transform'ttnons eust winch ml!
do the
In matrix notation
Q = Il ah
= p (6 ob)? where h = bp
6 must be chosen so that
; 0 0
0 a, 0
b ab = = [d,auj here &= { Ofor:

l for ' 75.7
_ J}
00 d..
IAXIMA AND QUADRATIC FORMS 369
Our problem is for the moment reduced to the simpler one of
determining the deniteness of a quadratic form which is a sum of
squares. Clearly this is positive denite, if and only if
,->0. (j=1,---,n) (24)
By the famous SylvestersLaw of Inertia it is known that the num
ber of plus and minus coefcients in the a's is independent of the
transformation employed. Clearly, therefore, the s, if all posi
tive after a particular transformation, are all positive for any
possible transformation.
Transformation into a sum of squares

Two out of the innity of possible transformations are of par
ticular interest since they show denitely what conditions upon
the matrix a are implied by its being positive denite. First, there
exists a transformation bgotten by an extension of the method
of completing the square"suchthat
a, = all
au 012
d _ (121 022 _ {If
2 an A1
au 012 an
021 022 023
_ 031 032 ass __ {

3 an an A2
an 022
(25)
an (112 ' ' ' alu

021 022 ' a2n
a,. = an]. 0112 an" "_ "

" au 012 a'lml A" !
Cim1.1 (ln1.2 an-lm-l

370 FOUNDATIONS 01' ECONOMIC ANALYSIS
It 15necessary and suicxent, therefore, that these ratios are all
posmve But
A1=au=Q(1.0. :0)>0
Hence, a necessary and suicnent condmon that a quadratic form
be posmve demte :s that the pnnmpal mmors (A, , A) must
be posntwe
Mzmmumpropane: of Q
Some preliminary observatxons are order before dlscussmg
exphcxtly the second transformatnon Because a quadratlc form
:s homogeneous of the second degree the hs 1f1ts Slgn ls always
posxtxveon the umt hypersphere dened by
hf + ha: + + hu= 1. (26)
1t 18posmve everywhere except at the ongln There IS no loss
m generahty therefore, m consndermg the sxgnof Q subJect to the
s1de condmon (26)
Smce Q 19everywhere coutmuous thh continuous derivatives
1t necessanly attains a mlmrnum on the closed hypersphere For
the form to be posmve ct 1s necessary and sument that the mzmmum
of Q on the mnt hypersphere be posztwe,because then, and only then,
IS1t posmve everywhere except at the ongm
This mnmmum can be calculated expllc1tly Cons1der
Q "= 2:: }?auhhi
Cf H Hancock Theoryof Mamma and Mamma pp BMI

Thu: 13to be seen as follows If Q were negatwe for any value of h( h) not
necessanly On the umt hypersphere nt would be negative {or some value on the hyper
sphere Suppose
X _ '
:
Then the pom: (fu/r , [i./r) :s on the unit hypersphere and
5 1
Q(;) "' ", Q)
ao that then sxgnsagree
subject to
?!
Zhi21=O.
l
Form the expression
H = ;? amhj M); h,.z_ 1),

According to the previous section it is necessary for a minimum that
2 ilah, 21m.-= o, (i = 1, ---,n)

or
xa.,.)h,- = o. (1:= 1, n) (27)
These are linear homogeneous equations which possess a non

vanishing solution if, and only if, the determinant
DO) = lao_ Mal = 0-6 (28)
This is an nth degree equation in Aand will possess n roots, some
of which may be repeated. For simplicity, we assume them to be
distinct. Let us indicate these by
M<hz< <)
At least one must be real since a real minimum does exist. Ac
tually, by a well-known theorem of Hermitian matrices of which
our symmetrical real matrix is a special Case, all are necessarily
real. Corresponding to any A," there exists a unique solution
(hf, - - -, hn")satisfying equations (27) and lying on the unit hyper
sphere dened by (26).
Also, multiplying the rst equation of (27) _by in", the second
by he", etc., and adding we nd
n
x. = >: " ahikhjk= w). (k = 1, n) (29)
1
It follows that the smallest of the As,)equals the minimum of

This is lled a secular equation, a delerminantal equation, and the characteristic
equationof the matrix 0. Its roots are the latent roots of the matrix a.
' H. Hancock, Theoryof Maxim and Minima, pp. 107109. See also the discussion
on orthogonality in footnote 9 below.
the quadratlc form Hence, a necessary and :uiczent condman that
a quadratzc form be posztwe demte es that all the latent roots of ds
mamx be paszme "
Similarly a form IS posntlve scmn demte If none of the )s's xs
negatxvc, although one or more xszero A form as negatne demte
2f, and only 2], all the )3: are negatwe
Second transformatzon mto a sum of squares
Thns can be proved stlll another way Consndcr the trans
formatton
1); = 2 11311. (1: = l, , n) (30)

I
It IS orthogonal, 1e .
>: :>: = 231:3.

n 1 (31)
]: .*]:." Il 51
1
' lt can be shown that A. ISthe mnmrnum olQ o l the set of pomt: lying on the umt
hypersphere and sausfymg the oonstramt def ned !
'n')" + n]! + + b,,I:. = 0
If we adJom still another constramt
131+ 59% + + 13.313.= 0
A. mammazesQ Generally x. mnmmazes Q subject to
Ein-=:
ZM=0 (,.-_| 71)

a
' In matrtx notation

b = [h "]
Orthogonahty :s dened by
bb = I therefore b = b
Therefore 9 = bl: 1mpl es h u b 9
To show that any no different 1columns are orthogonal note that
ahlt =- M! ah! = MI:
and that
0 = (hi) ahh -(Praia! -= (A; - h )(h) Iz
If the mo roots are unequal the very last product musuamsh and the h aare orthogonal
Because tuo vectors consastmg of Conjugate complex numbers cannot be 01111030al
this shows the truth of the prev1ousassert on that all roots of a symmetrtl mam!
must be real
MAXIM/1 AND QUADRATIC FORMS 373
Hence,
- h,- = $11,161, (: = 1, ---,n) (32)

and
(2(1))= z a- ?. ';hfpkhm
" 1|
ll
HIV]: 12 (12 aijhikhim)pkpmw
% we. (33)
This shows that the original quadratic form can be transformed

into a sum of squares whose coefcients are the latent roots of the
matrix a. This furnishes a second proof that the Xs must all be
positive if the quadratic form is positive denite, and conversely.
Characteristic equation of a irzatrix

Let us consider in more detail the nth degree determinantal
equation
DO) = lao" _ Mal =() _ >002 'K) (M 'M- (34)
Expanding out the determinant, this becomes

( X)" + c1(A)" + -+ cn_1(- A) + 6,, = 0, (35)
1Proof that the expression in parentheses above equals M is derived by multiplying

the ith equation of (27) by h.-"Iand summing. This gives
2 Z aiihfhs'" = M 2 113115
= Ab
1 l I
because of the orthogonality property of the transformation b.

" The above discussionimplicitly assumes that all latent roots are distinct. If they
are not, the matrix a can still, because of its symmetry and linear divisors, be transformed
into a diagonal matrix whose elements are the latent roots, but not by a unique orthogonal
transformation.
where
61= an + a + + a.: sum of all npnnczpalmmors
of the rst order
au au 011 au _ sum of all "(n 1)/2
62 =
au an an as: _ prmc1palmmorsof the
second order.
(36)
c, = sum of all n'/r'(n ~r) pnncxpal mmors of the rth

order
Ca= laul : D(0)

From the famlhar relatnonshxpbetween the roots and coefcientsof
a polynomxal we have
51=M+M+ +n=an+an+ +0"
0: = MM + MM + + MM
+ MM + + MN.
+ + Xl
(37}
c. = sum of all possxble products of (> , >)taken r

at a tlme
C.. = la! = D() = XIX: X.. > 0
Descartes' rule of sugns requn'es that all the c's be posxtwe If

all the X'sare to be posxtwe, and conversely It lSclearly suiment
that all pnncxpal mmors be posntwe
It ISobvxous that demteness of the quadratu: form Q xmplles
demteness of all quadratxc forms made up of subsets of the van
ables. for these are s1mply specml cases when some of the h's are
set equal to zero The latent roots of the matnces of all such
AIAXIMA AND QUADRATIC FORMS 375
subsets must be positive. From a consideration of the minimum
properties of the Asgiven in footnote 8, it follows that any such
latent root )8"satises the relation
xl ; v g A... (38)
From this fact we have another proof of the necessity that all
principal minors must be positive since the product of the latent
roots of a submatrix equals a corresponding principal minor.
Deniteness of inverse matrix
Repeatedly in economic problems it is important to be able to
evaluate the principal minors of the reciprocal or inverse matrix
a1 when a. is known to be denite.12 The following theorem is,
therefore, very useful. If a is the matrix of a positivedeniteform,
then a"1 is the matrix of a. positive denite form. Let
Q = hah.
Perform the transformation
h = a'lp, (39)
where a = a because of the symmetry of the matrix 0. Then
Q = p'craaf'lp = palIp = palp. (40)
This shows that the original quadratic form can be transformed
into one whose matrix is the inverse of the original matrix. From
our previous discussion this must be denite.
" An :42
A A
a1 =
:12 in."
A A
where A is the determinant of the matrix a, and Ar,-is the cofactor of the element in the
ith row and jth column of the determinant A = Iagil- a"1 is termed the inverse of a
beuse the transformation a" following the transformation a is equivalent to the
identity transformation I because of the property
ac = a"a = I.
Only non-singular matrices have inverses.
1Sylvester's Law of Latency states a more general relationship. The latent roots
of f(a) are equal to f(x) where the A'sare the latent roots of a. This followsfrom the
V. QUADRATICFoams Drmun. UNDER CONSTRAIN'rs
Let us now derive necessary and sufcient conditions that
quadratic form
is positive denite for values of h satisfying m( < n) linear cond

tions of constraint
fl
Z 54:41,= 0. (a = 1. , M) (41
1
where |g,l is of rank m. It would be possible of course t

eliminate m variables by use of the equations of constraint, sul
stitute in Q, and apply the theory of the previous section to
quadratic form in (n -m) variables. But this is lacking i
symmetry
followmg identities
ox = Mx
f(a)x = f(MJx
= f(UIx
Where f (a) = a". the theorem above emerges as a specualcase In the above theore
I must be a rational function
" It is, of course. suierent that a be the matrix of a posmve denite quadrat
form, since, If It 13posmve everywhere, It 15clearly posative for the subset of valu
satisfying the m conditions of constraint But this is not a necessary condmon Tl
failure to recognize this was responsible for techmml errors economusts' treatment \
utility theory up until Slutsky s classw treatment of the when
The mammum of
U = P'll . Il)
subject to the sndecondition

E par:= I
I.
does not require as secondary conditions that the matrix of the Hessran
[Wii]
be a negative denite form In fact, the demteness of this matrix ls no! invaria
under a general monotonic transformation of the utnhty mdex of the form
F = F0?)
F; = Fm
Fu : F.9d + F".,,
ms the correct conditions later to be developed are Invamnt.
M'AXIJHA AND QUADRATIC FORMS 377
. As in the previous section, there is no loss of generality in con
Sldenng the behavmr of Q on the unit hypersphere dened by
>?sz 1 = 0. (42)
This additional side condition is adjoined to the m equations of
(41), and Q is to be minimized. Form the expression
H = El: >; a.-,-h.-h,-

+ 2 ; u En:
l gihi M 1 hf _ 1). (43)
where A and. the ps are Lagrangean multipliers. At a minimum

we must have
m: Minn-kw zMfmo =0. = 1.) (44)

01'
((15,-_ tiylj + 1":gfa : 01 (i =. 1! ' ' '! 11)
ngh; = O, (a = 1, - - -, m) (45)
hf: ' 1 = 0.
1
A non-vanishing solution of these equations exists if, and only if,

the following bordered determinant vanishes,
as; X55; gi
D(X) =
gi 0
= (1)"A() _ >l)() _ )2) ' ' ' ( _ Xnm) (46)
where A > 0. The roots of this (n m)th degree equation are all
real because of the symmetry of the'determinant. Let them be
written in algebraic order
Mh-H lum
Then if the form is to be positive denite under the above con
straint, the smallest root, )easily shown to be equal to the abso
lute minimum of Q under constraint, must be positive, and con
versely. Hence, it is necessary and sucz'erztin order for a form to
378 FOUNDATIONS OF ECONOMIC ANAL}SIS
be pasztwe demte that all roots of the above bordered determmant
equahon be pomme
Recallmg that the roots are all real and applymg Descartes'
rule of Slgns, It can be shown as m the prevnous sectron that the
followmg bordered prmcxpal minors must all be of Sign ( 1)
(1,]: 1) ")
0 <(-1)"B' = (1)- m,gf
.. kn=1 ,m) MN
& 0 m+1rn
These (n -m) condmons are both necessary and sufc1ent
If the form 18to be negatwe derute under constramt, the de
termmants B' must alternate In Sign, the rst one bemg of sngn
( 1)", the second of opposnte atgn. etc , 1e ,
(1)'B'>0, m+1rn (48)
The mverse of the bordered mamx
It 15of consxderable Importance for many economxc problems
to know what IS 1mphed thh respect to the Inverse or rec1procal
of our matnx The latter can be wntten m matnx notatzon
ran ala 11 81"
R : 031
g ann
g gal
0 gum
0 __ [ga _g]
0 (49)
gl 3n" 0 0 J
The matrix has been part1t1oned mto four submatrlces, one 151:
by n, another m by n, stlll another n by m, and the last m by m
The matnx ISsymmetncal, 1e ,
R = R' (50)
The form xs negarwc demk 1f and only nf all are negauve If the root: (1Her
sngu It IS mdemte If one or more are zen: but the rest are of the same sagn xt :s
.MAXIJIIA AND QUADRATIC FORMS 379
The inverse or reciprocal matrix is dened by
RR1 = I = identity matrix.
It can be written
[Mud3H
g 0 ' q ' (51)
Performing the operation of matrix multiplication, we nd
[ad+gla+g9]=[f__0_]_I
gd+0 lg+0 0 I (52)
This implies the following four matrix identities:

ad +g' = I,
a + gq = 0.
(53)
g' = O,
g = I
It can be shown that is the matrix of a positive semi-denite form
provided Q is positive denite under constraint. It is only of rank
(n m) ; i.e., all determinants of higher order than (n -m) vanish.
MATHEMATICAL APPENDIX B
THE THEORY OF DIFFERENCE AND OTHER
FUNCTIONAL EQUATIONS
1 IN THE FIRSTmathematxcal appendlx I have Included a very
bnef summary of such theorems as play an Important role m eco
nomncanalys1s However, 1t seems desu'able to nge a fuller treat
ment of dlerence equations and other functIOnalequatxons Fxrst.
although ln many ways the theory of sumple difference equatlons
as more elementary than that of dtfferenttal equatxons xt is not
customanly taught ln mathematxcs courses, and no convement ex
pcsmons of the subject from the standpomt of the economxstare
to be found Second Just as dxfferentIalequations are of cardinal
Interest to the physxcnst,so dnterence equations are of the greatest
Importance to the dynamtc economist. and hence thelr exposxtlon
desenes a prominent place
A SmGLE EQUATION
2 A dl'erence equation can be written as an 1mphc1trelatlon
of the followmg form
fD'U + ") :'(! + n - I) .:) t] ==0 (1)
Tins W111be said to be of the nth order because nt mvo1ves the
variable y at n dlerent (equally spaced ') penods of time It may
be noted that the fact that the y yalues are dened for t values
differing by an mteger does not really Imply any restnctzon for
1fthe equally spaced values dlffer by any quantxty, we may redene
vin urne vana'oxe so as to make fms 'erence equal to nmty
Although the values of t dlffer by an Integer, lt 15not necessary
that I take on only Integral values However, for purposes of eco
' The econormst may consult wnth prot G Boole Treame on the Calculusof Funk
D!!! enccs (London 1880) L M Milne Thomson The Calculus of mete Dterences
(London 1933) P M Batchelder An Introdudmn :o Lmear Dzcrence Equation!
(Cambndge Massachusetts [927) N E Norlund Vorlewngen aber Dzcrenenrechmmt'
(Berlm 1924) for a more advanced treatment and blbhography
-DA
DIFFERENCE EQUATIONS 381
nomics it is usually only necessary to consider the values of y for
integral values of time. By breaking up our time into small enough
units we can keep the error involved in this procedure down to any
requisite level. As we shall see, this greatly simplies the mathe
matical theory and makes it resemble very closely the theory of
differential equations.
Equation (1) is an identity in t. The reader will see from this
that it may be written in very many ways, e.g., as a single relation
not between [:y(t + n), - -, y(t)], but as a single relation between
[y(t), - - -, y(t n)], etc. However, because equation (1) contains
as its last factor the explicit term t, rewriting our difference equa
tion in the manner suggested will involve a slight modication of
the form of the f function. Nonhistoricalsystems, which are of
the greatest interest in dynamic economics, do not involve time
explicitly and therefore omit the last I!in equation (1).
3. Difference equations receive their name because they involve
the value of a variable at different points of time. As written in
equation (1), however, there is no direct use made of the notion of
nite differences My, dened in books "'on interpolation as follows:
AW) = W + 1) - y(t),
NW) = AW + 1) - AW) = M! + 2) - N + 1)}
{W + 1) W)!
= W + 2) - 2W + 1) + W),
(2)
NW) = A"W + 1) A :'(!)
= Z ni')!z.!(])
(n- y(t +12 i)
Ay(t) = W)
It will be intuitively clear that a knowledge of n adjacent values
of y will permit us to calculate the value of y and of its rst (n 1)
differences. (See equation 6 below.) Therefore, by substitution
into equation (1) we can always write it in the following form:
FEMS), NIDU),' ' '! 43'): 30), t] = O, (3)
2Sec E. T. Whittaker and G. Robinson, The Calculus of Observations (3rd ed.,
London: Blackie, 1940), chaps. iiv.
382 FOUADATIONS OF ECONOVIC ANALYSIS
where F 15related tof as follows
PIA'y. . Ay.y. t] = ID + my
+Lzquy+ +A v tv+AyvyotJ (4)
Not only can every dn'erenceequation hke (I) be wntten as (3),
but also conversely, because of the relations gwen m (2), every
equation of form (3) can be ntten as form (1) Thus the mo
forms are completely equnvalent, and rye can choose bet een them
on the basxsof com emence Actually, the form of equatxon (1) xs
most frequently encountered m the hterature, although later It Will
be shown honr either may be handled
4 Thetr relattonshlp to each other may be further revealed 11
e mmme bney the notlon of an operator Thus, A may be
thought of as a dtfference operator dened by the rst equatnon
of (2) In a Slmxlarway he may dene the dlerentlal operator
D(= d/dt) by the relatxon
Dy(t) ==y'(t). DW) = Dy'G) = :v"(t). etc
In the last century the English mathematman Boole, followmgup
the earlier fruitful suggestion of Lelbmtz, employed the shnltmg
operator E dened as follows
EN) = W + l). EM!) = y(! + 2). etc
The successof the operator method stems from the recogmtxonthat
these may be treated as 2)"they ere algebra1c quantmes Thus,
leavmg out the 3: function completely. we may wnte out the
operator Identity
E = 1+ A (5)
Treatmg these as algebranml etpresszons subject to the usual b1
nomlal theorem we deme Immediately the relations
En=(1+A)*=1+nA+'1-(2_Qm+
n A
(??')
=(E1)=E"-nE""1+-E"2+ + '(6)
l
+(_ I) 09,15
(11 +
ate the agreement of the second of these 1th equatlon (2). and
the use of the rst of these m connectxonmth equatton (4)
Exercises:
1. The algebraic deviation of the price of hogs from the equilibrium value
is in any year %the deviation of the previous year, but of opposite sign. Set
up a difference equation for the price of hogs. Intuitively is the process stable
or unstable? (Hint: Use as your variable the deviation of price from the equi
librium value.) \Vrite out the rst few terms of the sequence.
2. Form a difference table of the numbers 2, 4, 8, 16, 32. How many
difference columns can you ll in with the given numbers? What is the general
rule relating the number of columns of the difference table which can be derived
from a given set of numbers? Note the triangular form,
3. Write the difference equation of the rst exercise in the form of equation
(3) rather than (1).
4. Write the difference equation y(t + 2) a(1 + B)y(t + 1) + ay(t) = 0
in terms of the E notation and also in terms of the A notation.
5. What is the meaning of E?
6. The receding difference operator V is dened by Vy(t) = y(t) y(t 1).
Give its relationship to E, corresponding to the relations of equation (S).
7. Write out the usual Taylors expansion for y(t + 1) in terms of y(t)
using the .D notation. Verify the relation 2 = E. (Hint: Use the series
expansion e = 1 + x/ 1! + x2/2! + ' ' '.) What would seem to be meant by
sine D? by 1/(1 A)?
8. What is the rst difference of ?? the second and third differences? What
is the third difference of any second degree polynomial? What is the nth
difference of any nth degree polynomial? the (n + 1)th difference of an nth
degree polynomial?
9. Write out the difference equation satised by the numbers in exercise 2.
10. Let us dene ascending factorial polynomials as follows: :d") = x(x 1)
- - -(x n + 1). Verify that Ax") = nx"'. Of what does this remind you?
SYSTEMS OF EQUATIONS
5. Up until now we have been concerned only with a single
variable. In most applications to an economic system many vari
ables are involved. Hence, we must introduce the notion of a
system of difference equations in m variables (yl, - - -, y). These
consist of a set of difference equations in the variables in question.
Just as a static system can only hope to be determinate if it consists
of as many equations as unknowns, a dynamic system must contain
as many independent, consistent equations as there are unknowns.
Thus, our system takes the form
Ey1(t+ n).y2(t + n). ---,ym(t + n);
3'1). 3'2): ' " 391.0);1 = O,
(7)
fill-310 + n)! y! + "')1' ', :Yn-(t+ 11);' ' ';

3/10)3'20). ', me);
384 FOUNDAHUN) U!"::,L'wkuu, {uv/11.191.)
As before tl desnred thts can be written m mte difference form by
makmg use of the relattonshtps between A and E It ts not neces
sary that each of the equattons m the above set should Involve the
same number of lags xt:s often convenient as a matter of notatton
to wnte each as If It contatned the same number of lags as the
greatest number n which occurs m any of the equations From
the dtscusswn at the end of section 2 of this appendm ntWill be
clear that we can shtft our ttme reference pomt so that some
y (t + n) appears m each equation
6 A difference equatxon system such as ts dened m equation
(7) can be thought of as an Identtty whtch holds for each value of !
It Is cquwalent then to an 1nntte number of 1mphcntrelations
Thts does not lead to overdetermmateness because each new time
penod denes as many new variables as there are equations In
fact Il wc conne ! to a mte number of consccutwe values we
shall always nd that ne have less equations than we have un
knowns Thus the system ISformally mdetermmate unless we ad
10mcertain so called tntttalcondtttons or boundarycondmons
This may be Illustrated by a sum of money gro mg at 5n: per
cent Interest per annum Its behavror over ttme follows a Slmplc
dtflerence equation but the process 15 completely unambtguous
only 11'we specxfy how much money there as to begm With or
alternatn ely If we specnly to what value the prmcrpal utll have
grown by any gwen date
In the above example of a rst order dt'erence equation nmolv
mg a smgle \anable only one tmttal condition can be arbttmrrly
speaed Therefore one 15sand to be the order of the system
lf ue had a 5}stem consnstmg of tuo dtfl'erence equattons each of
the thtrd order (: e each mvoh es three lags) then the number of
tntttal values which can be arbttranly prescribed ould be seen to
be tn o tunes three In general where there are m difference equa
trons 1nm variables and each mxolves n lags the number of arbl
trarrly assIgnable mtttal condlttons wdl be seen to be m txmes n
and this number wzllbe termed the mammum order of the system 3
7 Addmg tmttal condtttons equal to the order of the system m
effect suppltes us With the nussmg equations discussed m the pre
' In the ntercs .of expos t n || c nve ence a d u lo n ty I have departed from
tl e u ual mathemat cal custo n [say ng that n th 3 case the order of the system s a!
max!n t mes : By adopt ng the convent on descr bed at the end of sect on 5 l have
nsured that the order w ll be exactly 1: t mes m
vious section so that our unknowns no longer exceed the number of
our relations. Actually, it is well known that a system of implicit
equations equal to a specied number of variables may not de
termine a unique solution because of inconsistency, dependence,
multiplicity of roots, etc. However, in any relevant, empirical
economic system it is the task of the economist and not of the
mathematician to specify an adequate number of relations to make
the system determinate. Moreover, in causal dynamic systems it
is natural to think of causation as taking place unilaterally in time,
from past to present rather than from future to present. There
fore, it is convenient to concentrate upon the case where the speci
ed boundary conditions are truly initial conditions sufcient to
start the system off, after which it determines its own behavior at
each step. This means that the implicit equations of (7) can be
solved explicitly for values of gr.-(t+ n) as single-valued functions
of previous ys.
31,-(t+ n) = JEy1(t + n 1), ---,y(t + n -1);---;
y1(t), - - ,y(t);t]. (i = 1, - - -, m) (8)
Under these conditions, if we specify any initial conditions of
the form
35(0) y2(0) ' : _ y.,.(0)
, (9)
y1(n _ 1) y2(n _ 1) " ' y(n" 1)

then we can clearly determine from equations (8) the quantities
[yl(n), - - -, y,,.(n)]. Adding these to the previously prescribed
variables, we are in a position to determine the ys at the next
instant of time, and so forth ad innitum. In other words, by
the use of the familiar principle of mathematical induction we can
state an important existence and uniqueness theorem: a system of
dierence equations of the type indicated in (8), subject to prescribed
initial conditions as given in (9), possesses a solution for all positive
values of t. The solution is unique.
This theorem is valuable, not because it tells us how to nd a
convenient solution, but because it assures us that once we have
found a solution, and veried it by substitution in the difference
equation and initial conditions, then we can rest assured that it is
the only possible solution.
386 F0 UNDATIONS OF ECONOMIC ANALYSIS
8. In this section it will be shown that any system of order S,
whatever be the values of m and n, can be replaced by a system of
S rst order differenceequations in 8 variables. This reduction to
what may be called normal form is always possible by means of
redening S = mn new variables. Yu, - - , Y. as follows:
Y!. = y!
Y!: : Ey
. (i: 1!"Im) (10)
YM = Enly
In terms of the new variables the reader can show that the system
dened by (8) becomes
EYE = Y
E172. = Ya:
(i=l,'-.m) (Il)
Byte1.: = ni
BY,; = J.
Our initial conditions are written simply as
[YMOD = [Yd]. '1):|- (12)
By way of illustration let us consider a system consisting of
one second order equation in one variable.
N + 2) + a:y(t + 1) + my) = sin :; y(0) = bo, y(I) = b1. (13)
The order of this system is clearly two. Dening new variables
as follows
Y!) = ya)!
YA!) = 53(1). (14)
we can write the above system in the normal form
BY,) = _ (Y,) _ Onyx)+ sin !.
Elli) = Y!).
Yl(0) : bo.
(15)
Y:() = b].
The reader should work out several examples, such as that of
exercise 4, section 4 above.
It is to be noted that equations (7) can by a similar denition
of variables be written as S rst order implicit equations. Indeed,
an indeterminate system, involving more unknowns than equa
tions, can always be converted into an (equally indeterminate)
rst order system.
9. We have seen that a single high order equation can be trans
formed into a number of rst order equations. Is the reverse
possible? Can a rst order system (and, therefore, any system)
be transformed into a single equation, whose maximum order is
equal to the maximum order of the normal system? Usually the
answer is yes, although the reduction is easier to indicate than
actually to perform, and in singular cases, where certain conditions
on the partial derivatives of the fs fail, the reduction may be im
possible. Also, a well-determined explicit system may after reduc
tion be expressible only as an ambiguous implicit high order relation
in one variable. Thus, the well-determined system
By: = 3'12; Ey1 = yz + yz (15)
can be reduced to the well-determined second order equation in y1
Eyx - Eyn = 3:12, (17)
but in yz we get only the ambiguous implicit equation
(Ezyg Eyg y} 2 == 4y22Eyz. (18)
Let us suppose that we have a normal system of order S,
ya + 1) : JiEyl(t)v ' ! y-(t): t]! (i = 17 ' ' '1 S) (19)
or, in abbreviated notation,
EY = J (I, t).
It will necessarily involve ZS variables, namely the y's at two.in
stants of time. Given S equations in ZSvariables, it is notpossrble
to eliminate all but one of the ys and be left With a difference
equation involving only that y. If we write down the identities
E2Y = J(EY,t + 1)
EY = J(E2Y, t + 2) (20)
E'Y = J(E"1Y, t + S 1),

38B FUUNUAJJUIV.) Ul Luzvumxb ILIVJILJOIO
e shall have introduced as many new variables as we have equa
tions. But not all of these need be eliminated; only those involv
ing all variables except the one in question All together we
have S2 equations in S2 + S variables, and we seek to eliminate
(S 1)(S + 1) variables. Our number of equations exceeds this
by one, so we can hope to perform the elimination, the remaining
equation being the desired S order equation in a single variable.
10. Everything said about difference equations applies with
very little modication to dierential equations The reader
should write out a single differential equation corresponding to the
single difference equation (1) by writing out the latter using the
operator E, and then substituting for E the operator D. Simi
larly, he should write out the system of equations (7) in terms of E,
and then derive a system of differential equations by the substitu
tion of D for E. The same applies to initial conditions and to the
notion of the order of a system The argument showing how a
system of equations can aluays be reduced to rst order normal
form also applies directly to differential equations. He may check
his reasoning by referring to the rst chapter of the Moulton book
just Cited.
Exerczses
1 The national incomeof each of mo countries equals domestic imestment
plus domestic consumers' goods production plus exports The consumption
and imports of each country are respectively determinate functions of income
in that country in the preVious period Set up the system of difference equa
tions describing the behaVior of each national inc0me over time SpeCialize
these by assuming constant marginal propensnties to consume and import
2 Use the 45 Ime graphical method of chapter x to analyze whether or
not the equation, Y? + Y. = 9, denes a unique causal system If neces
sary, modify the definition of the equation slightly so as to remove all am
biguity Can the initial \aIue, Yo,be assigned completelyat pleasure (Hint
try a few arithmetical examples )
3 The system, Ey(t) = 2(t). 82(2) = 5 yft) + 2 3(3), was comerted to
normal form from a single equation in y What was the exact form of that
en,!mnnm ami what. demms. were.used. to.mangeant; it to normal. [mank
4 Given the system, Ey. = ya. Ey, = yi. derive a second order equation
in each of the variables alone
' The reader may consult F R Moulton, Dijermtial Equation: (New York, 1930).
pp 69.for a more complete discussiion of this problem taken Wlth specul reference to
differential equations For linear equations this is worked out in some detail in P A
Samuelson, A Method of Determining Explicnly the CoeffiCientsof the Characteristic
Equation," The Annals of Mathematical Statistics, XIII (1942), 424-429
5. A body falling in the earth's gravitational eld satises the equation,
(Fac/dt2= g. Convert into normal form.
6. Write out a system of di'erence equations in A form. Dene the con
cepts of order, normal form, etc.
11. If only numerical answers are desired, one may simply use
the difference equation at each instant of time to calculate suc
cessively the new values of the variables, and so forth for as long a
time interval as is desired. However, this may be tedious in prac
tice, and inevitably there must be rounding off at each stage so
that the accuracy of the solution deteriorates with its length. (For
a stable system if the errors of rounding off remain within a certain
limit, the errors of the solution will remain bounded.)
The method of actual numerical examples has had considerable
use in economics, and is valuable from the standpoint of providing
illustrations. Still one can never be sure that another arithmetical
example might not contradict the principles inferred from this one,
so that the method is of limited usefulness even for the purpose of
providing qualitative information concerning solutions. After all,
life is not long enough for us to complete even one sequence, nor
to try all possible initial conditions, nor to vary the form of the
structural relations.
Unfortunately, if we simply consider difference equations in all
generality, there is no simple analytical way to write out explicit
solutions. The same is true of differential equations. The be
ginning student may be taught a variety of ad hoctricks which may
work on particular equations, but the advanced student must rest
content with his existence theorem which tells him that there exists
a unique solution; often the proof of the existence theorem tells
him how he may (painfully) approximate to that solution.
In both of these lines much more complete results may be
achieved if we restrict ourselves to linear systems. That this may
be a serious limitation, especially for business cycle purposes, is
shown in chapter xi. However, in chapter x it is shown how the
assumption of linearity is not seriously limiting if we are primarily
interested in stability, especially in the small.
In much of what follows, therefore, we shall conne our atten
tion to systems which have two properties. First, they are to be
linear. There is the added advantage that most mathematical
E. Lundberg, Studies in the Theory of Economic Expansion (London, 1937).
390 FUUNUA'IIUIV.)
Ur [nu/"Lump annular.)
exposmons are concerned wrth tl ese Second they are not to an
volve tune exphcrtly Most economic systems of mterest are
causal rather than hlstonml m mture and satnsfy this require
ment thereby szmphfylng our task and enhancnng mathematlcal
tracttbrhty
In fact 1!we msust that a system be tnneless and at the same
t1mehne1r we are left wnth the exceedingly Simple case of constant
coeiczentr whose theory :s perfectly complete and elementary
We shall see that euctly the same thmg lStrue of differential equa
trons so that one who has mastered either of these branches has
httle drmculty Withthe other From a more fundamental pomt of
we each of these ISSimply the exemphcatlon of a more general
theory of llnear operattonal equations
12 There 15 however one nan lmear case n here a more or less
complete solutron can always be md'tted That Is the case of the
rst order equatlon of the form
y+1 =f(3':) (21)
The 45 diagram of clnpter *tpxctures the various types ol equn

hbnum solutions y. = a With the property a f(a) = 0 It IS
also clear from that dlagr'un that '1motion wlnch remams bounded
need not approach an cqlullbrlum posmon but may approach a
periodic motion (one of the boxes of the cob eb theorem) Some
of these boxes are stable and some are unstable We may test the
stabthty of a motion Wthh repeats xtself excry S perxods by the
method which Professor Leontnel has suggested Let us dene
f.('c) = f(x)
fe) = JU
(22)
ft) = ILL-105)]="f Ulf.-2(x) I] etc

It 18then truc that the solutlon of equatron (21) at any tune !
for gwen mmal condition yo ISgiven by
W) = Myo) (23)
W Leont el Verzogerte Angebotsanpassung und Part elles Cle chzew
7 lsdm fur Nalzonalvkanomie Band V (19.34)
However, this is more an indication of a solution than a solution
itself since the fs in question may be very complicated functions.
Suppose that (01,az, ' ' ', a,) represents a periodic solution of
the original equation. This will be true if, and only if, the new
difference equation
yt+1 = fa(yt) (24)
is such that
f,(a.,-) = ai. (1:= 1, ' ' ', S) (25)
The motion will be stable if, and only if, the above equilibrium
points of the new single difference equations are stable. This can
be tested by means of the methods developed in chapter x.
Exercise:
Show that the stability of the periodic motion depends upon
lf'(ax)f'(az) 'f'(as)l % l.
[Hintz What is f.'(a.-) equal to?]
13. The case of the rst order dierential equation which does
not involve time explicitly,
dy/dt = f(y). (26)
is one which admits of an immediate solution by quadrature, as
indicated in chapter x. Its richness of solution is limited, every
motion being monotonic. Hence, if bounded, the motion must
approach an equilibrium point. There may, of course, be a number
of equilibrium points alternately stable and unstable.
Exercises:
1. Let f(x) = x. What is the general expression forfdx)?
2. Show that every equilibrium value of f(x) is an equilibrium value of
f2(x). Is the converse true? What about the stability of the common equi
librium points, will they agree?
LINEAR SYSTEMS
14. A linear system is one which involves the dependent vari
ables in expressions of the rst degree, i.e., without higher powers,
cross-products, nonelementary functions, etc. For the purpose of
this section it is not necessary that time be excludedor that it enter
the problem linearly. We may write our linear system in the form
LDUH = f(t), (27)
392 ['UUJVUIIJJUJJ.)
ur Luvuuuzzu annular.)
where L stands for any general operation upon a whole system of
vanables and f represents a known functlon or set of functions
For the dlscussron of thlS section It IS not even necessary that L
represent dxerence equatlons rather than d1fferent1alequatlons
Rather ntmay represent an lmphclt functional of y over the whole
I Interval and In the notation of chapter m page 261 It could be
wntten as
LMD] ==PDG}: t]
The fundamental lmear prOperty which L must satrsfy [S as
follows
LU: + yz) = LCM)+ Lfyz) (28)
from WhIChfollows Immedrately the general relatlon
L(a1y1 + 02y: + + anyn)
= 01L(y|) + 02L(y2) + + anL(yn) (29)
From this general demt1on of a lmear system there follow 1m
medlately two Important general theorems
Superposmon theorem The salutwn of (27) when f 15made up
of the sum of two dz'erentfs :: equal to the sum of the solutzons op
propnate to each Mathematically thls may be wntten m sllghtly
more general form
Msn) = I lmphes HEM!) * ZaJ (30)
The proof follows immedrately from (29)
Thns means that the solution of a system wrth many forces
actmg on 1t IS equal to the sum of the separate effects of each
Even where tlns ISnot stnctly true the economlc sxtuatlon may
approxnnate to tlns condltron
For many lmear systems It may be posable to gwe the solution
for a fundamental set offs mto whrch any f can be decomposed
If thls ISthe case our problem 15Immedrately solvable
Our second theorem states The dz'erencebetween any two salu
tzons of (27) must satzsfy the reduced (or homogeneous) equatzon
LU) = 0 (31)
T1115follows Immediately If we substxtute each of the solutlons m
77) and then subtract one expressron from the other The right
hand function will cancel out, and according to (29) we may regroup
the terms on the left-hand side to achieve the desired result.
A most important corollary of this theorem is derived simply
by rearranging terms. The most general solution (i.e., any solution)
of (27) may be written as the sum of some one particular solution and
the general solution of the reduced equation (37).
This permits us to concentrate upon the more easily manageable
reduced equation in order to determine the uniqueness, multi
plicity, boundary conditions, etc., of the solutions. It is often
possible by ingenuity, good luck, or intuition to arrive at a par
ticular solution of the unreduced case, and the present corollary
tells us how to transform it into the general solution, if only we can
nd the general solution of the reduced case.
Exercises:
1. Except that there are no boundary conditions, show that the above
remarks apply to a single linear algebraic equation of the form, ax = b. Show
that when the reduced equation, ax = 0, has but a single solution, the full
equation has but a single solution. Can the converse be stated? Can you
state that the uniqueness of either equation guarantees the uniqueness of the
other? (Hint: Distinguish three possible cases, neither a nor b equal to zero.
both equal to zero, only one equal to zero.)
2. Consider two (or more) linear algebraic equations. Show that unique
ness of either the reduced or unreduced equations implies uniqueness of the
other. Show, however, that the reduced equation can have many solutions,
when the unreduced has none. Enumerate all the possibilities, and with the
aid of a textbook on algebra show how one can recognize each case. (Hint:
The terminology of matrices and determinants will be found useful.)
3. Show that the following integral, and mixed dierence-dierential
equations are respectively linear:
M + f; ku mane =f(t),
y'a) + ay(t 1) =f(t)
Do they involve time explicitly?
15. In view of the importance of the reduced or homogeneous

system of the form
My) = 0. (32)
we shall devote a number of sections to it. Most of our analysis
will apply to systems of equations, but occasionally it will be de
sirable to conne attention to a single high order equation in one
variable. Where difference equations are concerned, this may
, .
39.} ruuuua: .ru-.:v; uvvuv.u.v .......... __,
wntten m the form
ao(t)y(t + n) + an(t)y(t + n - 1) + + au(t)y(t) ==0 (33)
Much of the mathematncal theory of dt'erence equations IS
concerned thh the case where the a s are snnple polynomials or
analytic functlons Just as much of the classncaltheory of ordinary
differenttal equatlons ISconcerned With the case
WWW) + 010)?" 1(1)+ + an(t)y(t) ==0 (34)
where the a s are of the same Simple type
However 1fue msustthat the economlc system be nonhrstoncal
1e Independent of calendar tune and If we also msxst that It be
linear then the as must be constants and we have the case of
constant coefcxents
16 Whether or not t1me enters exphcntly 1t 18easy to derive
from equatlon (29) the theorem A lmear cambmatzon of a number
of solutzons of the reduced equatzon (32) zs ztself :; solutzan 1e
L(c;y1 + 0502+ + c y.) = 0 li L(y) = 0 (35)
Since the d1flerence between two SOlUthnS 15a lmear combma
tlon ue have the zmmedxatecorollary
u=o am
If the y stands for a whole set of variables the zero must of course
stand for a whole set of vanables equal to zero
Usually m the analysts of dxerentxal equations and dxference
equatxons as well the mathematlcran looks for a baszcset of spectal
solutions m terms of whmh any solutlon can be wntten by linear
combmatlon Thxslmear combmatlon wntten wnthundetermmed
coefments 15called the general solutzon of the reduced system
If there ISan exnstencetheorem whnchguarantees a umque solu
non of the reduced system satnslymg certam zmtxalcondttnons we
should of course expect the number of undetermmed constants m
the general solutxon to be equal to the number of arbltrarnly as
sngnable lnltlal or boundary conditions or equal to the order of
the system S
The reader should note that not every set of S distinct specnal
solutzons of the reduced system wnll form the baszs for a general
solutxon Thus If one or more of the gwen S solutlons can be
written identically in t as a linear combination of the others, the
set of particular solutions is said to be linearly dependentand cannot
form a basis for the general solution. For a linear combination of
all S solutions can in this case be regarded as a linear combination
of less than S linearly independent solutions; the number of con
stants in such an expression is necessarily less than S, and hence
we cannot hope to satisfy S arbitrary initial conditions.
17. However, the reverse is not necessarily true. If we nd S
solutions which are linearly independent (i.e., which are not linearly
dependent), it does not follow from our existence theorem that they
form a basis for the general solution. Indeed, there may exist no
general solution in the sense in which we have been using the word.
This is perhaps inadequately emphasized in the mathematical texts
because of their preoccupation with the special case of dierential
equations where a stronger theorem can be enunciated. Thus,
D2 '(t) = 0 (37)
has linearly independent solutions (1, t) which form a basis for a
general solution, in terms of which any solution can be written.
But
E2y(t) = 0 (38)
has two linearly independent solutions,
y1(t) = 1, t = 0; y.(t)= 1, t = 1.
m) = 0, z ; o; M) = 0, t .2.1, (39)
which do not provide such a basis. For the function
330) = 0, t g 2,
is a solution of (38) for appropriate initial conditions, and yet it
cannot be written as a linear combination of the two linearly
independent solutions.7
7l t may be objected that 330)does not satisfy the difference equation (33) for all
values of t, particularly for ! equal to zero. However, a solution is to be thought of as
satisfying a difference equation if for given arbitrary initial conditions it is in accord with
the prescribed law. To help clr up the present paradox, it should be pointed out that
solutions which are linearly independent on one interval of t may be linearly dependent
on another interval. It is also to be pointed out that the example cited is one in which
the solution of the difference equation is unique to the right of the initial conditions but
not to the left. It is also one in which the irreversibility in time disappears if the differ
ence equation is replaced by an equivalent" one of lower order.
396 FOUNDATIONS 0r ECOAOIHCANALISIS
The absence of a general solution does not me m that we cannot
nd the solutlon for gwen mmal condmons Our existence
theorem guarantees us tlnt ne can nd such '1solution Actually
such a solution can be \\ ntten ISa lmear combmatnon of basnc
solutxons However the aboxe evunple shows us that the linearly
Independent basic solutions may haxe to be dt'erent sets for every
mltnal pomt there need not cust a set of basxc solutions which wrlI
serve umformly for all mltral times
LINEAR OPERATOREQUArrows wmr Covsmvr
Corrrlcrrms
18 I turn now to the case of drfference or ddl'ercntnl cquatrons
With constant coefcrents Since xt ml! be shoxm later that the
solution for equations of any order can be written terms of
solutions of the rst order C'lse I shall consrder thus Simplest of all
cases In the rst order case our drfference equation becomes
(E mm = K!)
yUo) = b (4)
and the correspondmg differential equatlon becomes
(D - 9M!) = f(t)
m =b 42)
Each of these may be consulcred as specral exemphcanons of
a general rst order equatron m a Simple opentor 11
(h " 9)y(f) = f (f)
?(10) = b (43)
For h to be an admzsszbleoperator nt nsonly necessary that there
be an emstence theorem guaranteemg that the above equation has
a umque solutnon for the arbltranly specred Jamal condition
How thrs exrstence theorem rs estabhshed need not concern us at
this pomt we hu e seen how it can be done for differenceequations
and the reader may refer to any orL on dl'erentnl equations for
the corresponding proof In addntron to E and D examples of h
are A 0 E + 5 etc On the other hand such an operator as D
would not be an admrssrble h because the system correspondmg
to (43) would not have a umque solution Thus :s because mo
lmtIaI conditions are necessary rather than one to satlsfy a second
order dxfferentlal equation
.We may . suppose the solution of the above system to be
written as a linear functional of f, with q and b as parameters, or as
F(t, g, b; f)where F has the fundamental linear property
FU, g, Clbl + ' ' + cabs;lel + ' ' ' + csfs)h
= GIF! q, bl;fl)h + ' ' ' + CSF(tyq: b3;f8)h' (44)
From this fact it is easy to verify that h is a linear operator
with the properties
[t(Clyl + 62372)= Clhyl + Call,
Iz2 = lz(hy),
' (45)
h"y = hh"y,
h"! = ly,
110 = 0
P(/1)Q(h)y = Q(h)P(/t)y.
where P and Q are polynomials.
19. So far we have not dened It raised to a negative power.
It would be natural to dene
y = hf(t)
as the solution of
by = K!)
However, it is clear from our existence theorem that this cannot
have an unambiguous meaning unless we specify some initial con
dition. This corresponds to the familiar indeterminacy of the in
denite integral of a function, Dlf. We need a constant of inte
gration or a denite lower limit of integration before there can be
a determinate answer. We shall not have occasion to use negative
powers of It even though readers familiar with the Heaviside opera
tional methods will realize the mnemonic advantages to be derived
from such a procedure. These advantages are counterbalanced by
the aura of mystery which surrounds some of the usual expositions
of this method. In general, the methods here outlined will be of
classical type, albeit in somewhat amplied form. It is the
present writer's conviction, which need not be sustained here, that
a careful examination of the Heaviside-Cauchy operational calcu
lus, of Bron-m1chWagner contour Integrals, of Laplace transforms,
etc , Wlllshow that where these dlfer from the classxcal methods
the advantages are m favor of the latter These remarks are
tended to apply to ordmary dt'erenttal equattons of mte order
Wlthconstant coccxentsand do not refer to the use of the Laplace
transform 1n connectxon w1th partlal dt'erentlal equatlons, nor m
connection w1th Integral equations of faltung type, etc
20 chg rst concerned wnththe reduced equatlon, wheref(t)
vamshes we concentrate upon solutions of the form Fa, q b, 0)
Because of the hnear prOpertIes of h gwen m (45), this can bc
ntten m the special form bF(t, 9 1,0). Hence, pecuhar :mpor
tance attaches to the solutnon of the reduced rst order equation
Wlth mutual condntlon equal to umty Let us dctcrmmc, there
fore, the exact form of the F function for h equal to E and for 1;
equal to D
Clearly t(
Ey = ay. We) = I.
then
We + 1) = Wo) = q.
(46)
We + t) ==quo +t -1) = qqyuo +1 2) = = 9.

:) = .
Ol'
F: q: 15( = {l"' (: t)
Of course, the nal proof of the above relatlon IS to be found
m the fact that when substituted mto the difference equation.
q" satises 1t and the boundary condttton ln questton Our
existence theorem guarantees that It 15the only solutlon
We may rewnte the dlerentxal equation (42) the form
dy/y = gdt (4?)
Integrating each Slde.talung anttlogs and msertlng the appropriate
constant of Integratlon so as to satlsfy the Imtlal condntlonof umty
at t equal to to. we easxly nd the solutxon to be
F0, q. 1.0)., = 0" > (: ==>to) (48)
' G Doetch Thorie und Anwendung de: Laplace Transformazzan (Berlm 1937)
The reader may verify that the operators A and 2D have re
spectively the fundamental solutions
F0, q, 1; 0)A = (1 + Q)H: ( lo) (49)
and
FU! q, 1i O)) : (t/l0)q' (1 - to) (50)
Thus far, it has been implicitly assumed that 9 is a real vari

able. However, no detail of the procedure or proof would be es
sentially changed if g were assumed to be a complex number of the
form (a + ib). It may be argued that no real economic process
need be described in terms of complex numbers. This can be
readily admitted. However, when we come to equations of higher
order, it will be shown that the solution can always be written in
terms of purely algebraic combinations of the roots of higher order
polynomials. In order that every polynomial have exactly as
many roots as its degree, it is necessary to introduce complex
numbers. Because these occur in conjugate pairs, it will always
be possible to regroup terms at the end and express everything in
terms of real numbers. But it is most convenient to work with
complex numbers in the intermediate steps.
21. We may now inquire into the stability of this fundamental
rst order solution. It is clear that the initial condition enters only
as a scale factor and cannot affect the problem of stability. Under
what conditions will
lim F(t, q, 1;O);, = O? (51)
locn
If we examine each of the four specied Fs corresponding to the

four distinct operators, E, D, A, tD, it is clear that we have the
following respective conditions:
E: absolute value of root must be less than one,
or lgl < 1;
D: real part of 9 must be negative, or 12(9) < 0;
A: absolute value of (1 + q) mnst be less than one,
01' II + (.Il< 1;
tD: real part of 9 must be negative, or R(g) < O.
In the complex plane the rst of these is interpreted to mean
that q must be inside the unit circle so that its absolute value or
modulus wrll be less than unlty For the dlfferentlal system the
value of 9 must be m the left half of the complex plane Later 1t
wxllbe shown that the stablltty of the reduCedequatton detetmlnes
the stabllnty of the system
22 Havmg mtroduced a basm solutlon of the reduced equatron,
FU, q I. O), I now mtroduce a basrc specral solutlon of the un
reduced equatron By combmmg these two. the solutton to the
general rst order system (43) can be wntten Our specml solution
of the unredueed equatton rs that for wlnch the matra! condstron,
b, rs zero In accordance wnth our above notation thlS can be
written as FG q, 0 f ):. How xt 15to be found for any given oper
ator need not concern us here, our emstence theorem assures us
that xt emsts. and often suggests how xt xsto be found
The reader can verxfy that the followmg do represent basxc
speetal solutrons for the respectwe sampleoperators
D ""' e9"*')f(a)da,
J; (52)
ott
E 2 9f(z)
etc
Because of our enstence theorem, no proof other than verication

xs necessary
Consxder now the ewpressnon
1) : F, gi {)th + bF, Qt1,0)1. (53)
Substltutron shows that It rs a solutlon of the general equatton (43)
Its mntlal value IS
yU) : F!grow): + (71:00.9 1|0)h : 0 + bv (54)

satisfymg the correct mmal condition, and thereby constntutmg
the umque solutlon
23 We shall show that the most general system of any order
has a solutlon conSIstrngof two parts a specxal particular solutxon
of the reduced equatlon wrth vanlshmg 1mt1alcondmons and a
general" solution of the reduced equatton thh asagnable con
stants suFcrentto satlsfy the prescnbed m1t1alconditions
In fact, once the rst order case has been thoroughly analyzed,
the general case yields to a purely algebraic treatment. In what
follows, I shall try to bring to a completion the Heaviside program
of algebrizing the treatment of simple operators. As far as [
know, the mathematical literature, because of its principal pre
occupation with special operators, stops short of such a complete
program; at some stage the concept of integration or summation is
introduced into the discussion. Moreover, the present fashion of
thought introduces transcendental notions such as the Laplace
transform even in the simplest cases of ordinary differential equa
tions with constant coefcients. It will appear from what follows
that the Laplace transform, however useful it may be to one who
has grown accustomed to it, is neither necessary for rigor nor
necessary for analytical and computational short cuts. In a
fundamental sense it is an artice which cancels out of the prob
lem, and aesthetically it represents an alien intrusion having no
natural" connection with the problem under discussion. Further
more, the present treatment may throw light on the connection
between so-called classical and operational methods, showing
their complete equivalence and identity. Aside from the lament
able aura of mystery surrounding many discussions of operational
methods, the literature abounds with diverse, contradictory claims
and counterclaims as to the advantages of the alternative methods.
Despite the facts that simple operators are being discussed and
that an algebraic treatment is being used, the present treatment is
essentially classical, with one important addition: full use is made
of the specialized algebraic character of the equations determining
solution constants. When so amplied, the classical methods not
only arrive at the same results as the operational methods, but
actually go through the same intermediate steps. Wherever the
two methods diverge, the advantage lies with the classical. For
those who must have their Laplace transform, the concept is suit
ably generalized to any operator in section 29.
24. We may begin by enunciating what is a theorem and not
an assumption: If h is a simple operator such that the system
(h - q)y =f(t), (t a to)
you) = b (55
has a unique solution for suitably limited functions f, and for 9 an
arbztrary complex number, then the general nth order equatwn
P (th = kw + euh"? + deb"? + + anxhy
+ any= fa),
300) = b0!
hy(t):]z=zo = bt.
(56)
h"y(t)]-r. = b,
has a solutron. and the solutlon nsumquc
To prove the rst part, I shall specxfyone solutlon, leavmg lts
umqueness to later drscussxon FllSt,let us note the algebratc fact
that every nth degree polynomml has exactly n roots If we count
each mth Its approprmte multzpllcnty If q. denotes the 2th root,
we may nte our polynomlal m the form
Pn(x) = (x _ 1)(x _ g!) (x "' a) (57)
Let us dene 1:new polynomials as follows
P;(x) = P,.(x)/(x - q,) = ZG'xa:_

n~2
Pu2(x) : Pu-l(x)/(x _ Q:) = : a Wxnz"
, (58)
Pf(x) = P?+\(x)/(x _ gran"): ? au rxrt'
mc) = P2(x)/(x g) = za. Ixh = x g
P005) = P,(x)/(x _ 9n) = 1

' For brevuy we shall \wne ($30)] - , as hyUo) where there 1.3no posSLbletyOf
confusion
Also, let us dene some new functions as follows:
0 " 91)Z1(t) = Z00).

(1 * 92)Z2(1) = Z1(t),
' (59)
(h g,.)ZnU) = Z,(t).
By' our hypothesis on a rst order system, each of these is deter
minable once we specify initial values [Z1(to), Z200), - - -, Z,.(to)].
In a moment these will be prescribed.
By virtue of the denitions of Zk(t) in (59), it will be clear that
f (t) = ZoU)
= (h _ 91)Z1(i)
= (h ' 91)(llgaza) (60)
= (h gl) (h -qn)Zn(t).
Thus, Z,.(t) is a solution of our equation, and it remains only to
make certain that we pick the Z{(to)properly so as to yield the right
initial conditions.
It is easily veried that
Znj(t) = P,-(h)Z,.(t). (61)
Evaluating this expression at the initial time we have
a;_,h""Z,,(to)= ai,jbji : zit1'00)

0 0
(j=0,1,2,-,n1) (62)
These are sufcient equations to determine the right-hand terms
for a given set of bs.
To recapitulate our construction: nd the roots of the poly
nomial P ; by synthetic division or otherwise dene the n new
polynomials, Pi; evaluate Z,-(to)in terms of the bs; by repeated use
of the methods of the last section, equation (54), solve in turn for
Zi(t). 220), - ~,Zn(t), using the initial conditions just determined;
the resulting Z,.(t) is the required solution. This Proves the rst
part of our theorem.
404. FOUNDATIONS OF ECONOMIC ANALYSIS
To prove uniqueness e shall assume that there are two dls
tmct solutions mdshow that this leads to a contradictlon Clearly
the dl'erence between any tuo solutions of (56) ml] satisfy the
reduced equation wrth :nrtnal conditions all equal to zero We
have only to show that the reduced equatron wrth zero rmtral con
drtrons cannot have a solution whrch usnot :dcntlcally zero For
the rst order case our existence theorem msures that there can
be no other solutron than zero We shall proceed by mductlon to
show that 1f'1reduced equatlon of order (n - 1) has a umque zero
solution [or zero rmtral condltxons then so must an nth order
equatlon
Let us suppose that the (n -1)th order equation .has a umquc
vamshmg solutron and the nth order has a non vamshmg solutlon
Y(t) Then
{PUYl = (h _ &)le uUYl= (h _ quz = 0 (63)
where
hY(tg)=0 (r=1 nl)
YU) # 0
and Z rs dened by thrs equatron Our hypothesrs on the (n 1)th
order equatlon s umqueness tells us that Z must be nonvamshmg
smce otherwrse our nonvamshmg Y would satxsfy the (n 1)th
order equatnon On the other hand the zero values of the n mrtral
condntrons of Y make Z(tu) be zero Wrth 'un 1n1t1alvalue of zero
and satlsl'ymg a rst order equation Z must vamsh Identically
Thus we hate 3.contradnctnon and our conclusron concermng the
uniqueness of our solution rs proved
It ls to be noted that the case of repeated roots m the character
lStICpolynomial P Introduces no dlfcultxes 1n the problem Also
It 15to be observed that the method of nding a solutlon outhned
above ISnot necessanly a good one to follow m practical numerical
computatron However smce xtdepends only upon repeated solu
tron of rst order equations rt can be readily mechamzed or reduced
to a routme ll some surtable method of solvmg rst order equations
18at hand (e g dterentral analyzer punch card equrpment etc.)
Etam ses
1 Wnte out the soluuon for a reduced second order (I ercntral equat on
With constant coefc ents wh ch has y(0) = l y (0) 0 vh ch has y(0) = 0
1'(0) = l
DIFFERENCE EQUATI ONS 405
2. Write out the solution of the unreduced equation with vanishing initial
conditions.
3. Repeat (1) and (2) for any operator h, using the F function.
25. Having established the existence and uniqueness of the

solution of the general Operator equation of the nth order with
prescribed initial conditions, we may explore better ways of ar
riving at its solution. Our work may be divided into two parts:
nding a particular solution with vanishing initial conditions, and
nding a solution of the reduced equation which satises the pre
scribed initial conditions. Clearly, there are other ways of splitting
up the problem according to the properties the particular solution
is to have. Thus, in many physics problems the right-hand func
tion, f(t), represents a periodic driving force, and the particular
solution sought is the so-called steady state response of the system,
while the reduced equation solution yields the transient, which
dies out in a damped system.
The reduced equation involves no arbitrary functions and is
simpler. We discuss it rst. Its solution for the most general
case can be built up out of rst order solutions, F(t, q, 1; 0) For
brevity we shall write the latter as F(t, 9). Some of its properties
may now be developed.
First, since
Fo. q) E 1. (64)
ah _
5? Fan, 9) = 0- (65)
Also, by its denition as a solution of the reduced rst order
equation,
hF = qF,
th = {T,
- (66)
lz"F = q"F,
and
P(h)F = P(q)F. (67)
Because of the linearity of our system, we have
2,, Pht,2Pht,1)
P(h)[y(t.4q)2-:l(19)]= ()y( gaiq} ):v( 9 _ (68)
If y is a continuously differentiable function of the parameter q,
we find in the limit
m) 3%= 53%) (69)

For the moment we disregard all boundary conditions and seek
solutions of the reduced equation
P(h)y = 0.
First we establish a lemma: ll q is a root of multiplicity s ol
P(q) = 0.
so that
P(q) = 0 . .. .
PC) 75 0, (I " 0:11 ' ' vs " 1) (70)
then
pagimmao. (k=0,l.--,s-1) (71)

&
where the last expression is short for [P(h) 5%;F(,g)],,_;

Actually
6" "
PO!) 597F = a? PUFU. (z) = 53: [P()F(l. t2)]
= mom. c:)+ kP'() (%Fa. s?)+ ~

+ Pa) 36;
Fu. a)
= 0, (k < 3) (72)
term for term because of (70).
Of course, any linear combinations of special solutions of this
form are also solutions. Every polynomial has n roots, when rc
peated roots are counted according to their multiplicity; and every
distinct root yields as many special solutions of the form given in
(71) as the multiplicity of the root. It followsthat we have
'
Em.=n
l
special solutions, not yet shown to be all linearly independent.
We take the most general linear combination of these
a nul
y = C10F(t,91) + Cll'FU, 91) "i" ' + C1.m,1'agm_l_1
FU) gl) +
a amz-l
C20FU. 92) + C2152 FU, 92) + ' ' ' + C2,m,1
gg? FU, 92) +
...................................................
ooooooooooooooooooooooooooooooooooooooooooooooooooo
a am,1
CrOFU,gr) + cf] FU; gr) + ' ' ' + Cram15E: FU, qr)
r mkl ai
= i=1
Z 1=o
Cl:.i',-
g 170,91). (73)
where (gl, gg, - - -, q,) are the distinct roots of P with respective
multiplicities (ml, m2, - - -, m,). This linear combination is a solu
tion of the reduced equation. It remains to determine the n co
efcients (Cm, - - -, C) so as to satisfy the n arbitrary initial
conditions in (56). Thus
. f inkl a). _
ht:l=l 1:0
z ijqu(t:Qk)=biy
8 (1:0,1,"',n_1) (74)
or because of (66) and (69)
r ml J'
ZMWw(QkWW(W
t:! j=0
From (64), F(to, g) is unity regardless of g, so we have

"Ik1
S: ij(q;_.")(" = b... (a = 0,1, ---,1z 1) (76)
k=1 j=
where
dj . . il __. . .
. = .* (j) = -.-. ;' J, ?:
[dg] g :lq-qg (9L) (1_ J)! gl ( ])
=a <p an
408 FOUADATIONS OF ECOAOIHC ANALYSIS
When all of the roots are (hstmct the determmant of tlns
system takes the famxhar Vandermondc form
1 l l
Q: : 9n
m q2 9
= II ((P ... Q:) (78)
:>;
qlnl gzn-I gnu-l
hell does not vanish Consequently the n equations can be

solved to yield the n C s so as to satlsfy the n mltlal condxtlons
The proof of the Identlty (78) follows from the fact that the de
termmant 15a polynormal of the (n 1)th degree m terms of any
g., and vamshes for any 9, equal to q. Thls plus the essentnal
symmetry nn terms of the qs leads to the rzght hand sude of (78)
Au alternative proof would use Induction
In the general case of repeated roots the determinant becomes
1 0 o t o 1 0
q: l 0 0 cv 0
g zq. o q,: 0 q 0
= (79
91 (qt) (W)" " 9: (q: JW 4} (M"
gt" ! (?!IMC", (q' I) I": II 4? I (qu 1] Ir! qu I (qq I).: I)
no mo q s bemg equal Thns generalzzed Vandermonde deterrm

nant can be shown not to vanlsh ' This may be proved by 1nduc
tron but for the present purpose we may omlt the proof of this
fact, and sxmply note 1ts consequence all of our constants can be
uniquely determined
Here I should hke to pomt out that the mxcrsxonof a general
rzed Vandermonde matnx can be done once and for all, and the
result 15apphcable against \arymg mltxal CODdlthllS The exact
form of the mxersmn xs expedrted by synthetic d1v1sron and as
closely related to contour Integr'mons and the expansxonof a poly
' P A Samuelson A Sample Method of Interpolation Praceedmgs of the banana!
Academyof Sama: XXIX (1943) 397301 AlsoF R Moulton Dacrmlzal Equa
1: (New York 1930) chap xv
nomial in partial fractions, so that the classical and operational
methods meet."
26. I turn now to the problem of determining the unredueed
special solutions with vanishing initial conditions. In the proof
of our fundamental existence theorem we have seen one way that
this can be built up out of the basic solution of the rst order
equation F(t, q, 0; f);,. Once the latter is tabulated or catalogued
for a great variety of the fs that arise in practice, the remaining
work is simply that of calculating
F(t. qi, 0;f);. = Z1,
F(t) gl, 92, Oif)h = FU: Q2.O;Z1)h : Z2)
F(t,q1.q2,q3,0;f)h F(t.Qa,0;Z2)h = Z3
(80)
FU! gl! QZ, ' ' '2 gm Oif)h FU! gm 0; Zn-I)h = Zn.
The last is the required solution.
In this section an alternative procedure is described whereby
our Specialsolution with vanishing initial conditions is built up out
of the rst order solution, F(t, g, 0;f);.. First, let us recall the
algebraic fact that any rational expression of the form
1 = 1
($112, = n) (81)
P(x) (x _ Q:)m
!=]
can be split up into partial fractions of the form

__1__ Au A12 Jan..
P05) _ (x _ 91) + (x 'QI)2+ + (x "' %),"l
__42_ . Jill_ ._""_'_._. 82
+(x_g2)+ +(xQr)+ +(x_qr)mr ( )
In the usual case where all the roots are distinct, the coefcients
are simply 1/P(q.), respectively. When we have a repeated root,
the coefcients are given by
1
A" : Wim-mr:d'" ELLE]
P(x) ==a. (83)
P. A. Samuelson, "Efcient Computation of the latent Vectors of a Matrix,"
Proceedings of the National Academy of Sciences, XXIX (1943). 393-397
DIFFERENCE EQUATIONS 4II
problem. After substitution into the nth order equation we have
P<h>y 1=l
(:1 z A..-[5503]
' QQ (h gain-(z. g. on.
_ r m; P(h) 90
* EaiAjl(hf'qiylf ( )
UU)
by virtuc of the identity (84).
Note that our discussion of negative operators was purely
heuristic and not germane to our proof. The expressionin brackets
does not involve negative operators although notationally it appears
to do so.
This completes our second construction of the unreduced equa
tion with vanishing initial conditions.
27. The elementary mathematical literature on differential
equations abounds with trick methods for arriving at particular
solutions of the unreduced equation for special f's (e.g., poly
nomials, exponentials, trigonometric functions, etc.). At this
point it is only necessary to discuss two such methods.
The rst is a formal series expansion, of even greater importance
in the eld of difference equations than differential equations where
convergence is not always certain. Let us write the formal
expression
1 1 _ . 2 ..
P_(x)=ao+a1x+ +anxno'l'lx'l'zx + '
Multiply through by P(x) and equate coefcientsof like terms to

get equations determining the innite sequence of B's in terms of
the as. .
1
60 = ;; ,
_ (05150)
51 _- Oto !
_ (01151 + 63250)
52 -a , (92)
B = _ (a16,_.1 + (12632
ao+ ' + (1.30) _
412 FOUNDATIONS OF LCOIVOMICANA! }SIS
In pnctxcc there are & mncty of we. of determmmg the 5 se
quence from the 5 (algebraxc dlvzsmn, tltf'ferenee equations
Taylor's exp1ns1onof I/P(x), dctcmumnts, etc)
Forma). a mrtlcuhr solution of
P(h)y = !
ISgncn b) the mmte series
l " ,
= F05! - EONI! (93)
u.here l/P nsno dened differently from (86) and (67) Whether
there ISany mcmmg to the right h'md mic thtt IS hether the
sencs 15dened 1nd comm-gent or not depends upon the h, the I.
and the a's Ilms
(mo + I)? -"-f (94)
has 1 nite com crgcnt solutlon for f eqml to t of the form
y = If rqu + mD'f amv +

= t - (1121+ al'l! n-1 + 0 -0 +
+ ( 00"! (95)
While for] equal to c e Inu. a forum]scan
) = % ("' a.)'Dc
= e S': ( mx) . (96)

0
which com crgcs far INN lcqs than one, but teems to ducrge
othermse Hoxxexer.if we use the "fornnl" adentltv
Z0 (" mk) '"Tilaa . (97)

we get the correct ansuer, c(l + mx)" although not '11ays b)
a.hammam amb.
Let us write a difference eqmtzon m the {arm
day) + ar:y(t- l) + + cay n) = P(E")y(l) =f(t).
EMO) = )(0) = 0.
I:"y(0) = 0. (98)
Enl)y(0) = ()
f(z) : 0, (: < 1)
Then
] no
y(t) = mt) = B;E1f(t). (99)

This willalways terminate aftert terms, and willsatisfy the equation
and initial conditions.
Exercises:
1. Show that B. is a solution of the equation
P(E")y
y(0) =
_ 6110?)
r + 0tny - 1) + ' - ' + 0100 n) = 0,
3'( 1) = .
y(- n + 1) = 0.
2. If P(:c) has n simple roots, gr, and R(x) = ao"x"P(x), show that
" 1
: ___ . (FIn
B. 21;Rm <q.) .
3. If a. equals a/t!, what is B,? (Hint: Consider series expansion of e
and 1/e.)
4. Write B, in determinantal form.
5. Write out the rst three Bs in terms of ds.
28. The second artice for arriving at a particular solution

comes into use when f (t) is itself the solution of a reduced rst order
equation. That is
P(h)y = f
where
hf = qf. (100)
We try a solution in the form
y = Cf, (101)
where 6 is a constant. We have after substitution and cancellation
off
cP(q) = 1. (102)
If q is not a root of P, this has a solution,
_ _1_. 103
" P<q> ( )
414 FOUNDATIOAS OF ECOAO'IIIC ANALISIS
In the Important clectnml engxncenng studies of frequency re
sponse, c ns the transfer function, for q a pure 1m1g1nary, at 15an
impedance
If 57ISa root of P, e hmc the phenomenon of resonance, and
we must try for our solution
_ c (TVU)
y ,,q_. (104)
here m + 1 rs the mttltupllcnty of the root 5'; After substitutaon

and cancellanon c nd
1
c P"(q) (105)
If f 15composed of tuo or more components, f,. f:, , Inch

are each solutions of the rst order reduced equatzon mth g equal
to 91,9; , by superposztron 0 nd a spccnl solution of the
unreduecd equation to be
1 l
y = 33ml} + m]: + 5106)
If any of the 9's are roots of P, the modxcatlon gn en m (104) ml!
of course bc m order
Important enmples are prox Ide by [)CI'lOdlC dmmg forces rn
the eld of d1fl'erentmlor difference equations Any ell behaved
pernodrcfunction can be n rrttcn as a Fcurler serres, of Inch each
term satises a rst order dn'erentnl (difference) equation mth
pure lm'lgmnr) (comple\ root of umt modulus) coefczcnt Unless
perfect resonance occurs, as 1t cannot m a dmtped system. OU!
solutlon can be bullt up term by term of the same harmomc ele
ments but w1th coelecrents dzermg by the factor, I/P(q.) The
resulting solution (nf 1t converges) ml] be a periodic functwn of
trmc of the same penod as the dm mg force
Fmally, if our [(I) can be written as a general mtegral
fu) = f meme). (107)

where
hm. q) = 9%. 9).
and 6 represents some contour in the complex plane, then we try
the formal solution
y = f mdm. (108)
In cases where P(q) vanishes, this may be an improper integral
whose limit nevertheless often provides a satisfactory solution.
The mathematician will recognize that in the eld of differential
equations, all the fs that are likely to arise in practice can be
written in the form (107), by setting q equal to ia, and F equal to
ef, and using Fouriers integral theorem.
29. The artice of the Laplace transform will now be general
ized to a wide class of simple operators other than D. The Laplace
transform is dened as
(q; y) = f: ey(a)da, (109)

which will be convergent for a wide class of ys, for the real part
of g less than some constant.
The fundamental property of the Laplace transform which
makes it a useful artice indifferential equations is as follows:
(q; Dy) = 9:601:30 y(0).
(q; D2y) = q(q: y) qy(0) - Dy(0),
(110)
43(9:030 = 98(9; DHy) D"y(0)

= q"(q:y) qy(0) q""Dy(0)
- qD"y(0) - D"y(0)
For then taking the Laplace transform of both sides of
P D = ? (L,-D' = t ,
( )y y f() 1)
Dky(0)=bk| (k=0,1,",1Z1)
weget
P(1)(q; y) = 5301;!) + l?e(aogF'1+ mgr2+- ~- + a)
+ wow-2 +- -+) +- -
+ bn2(aoq+ al) + bnla'o- (112)
Thrs ISan expresswn n q and involves no operators Solvmg
for the expressxon Involvmg the unknown function, we have
(113)
(q.y)= Eg+ Clg%;)
= P(st).
where R 15a polynomml 1m olvmg the mutual condltxons, 6;, and
ISless than n 1n degree Because I IS known, the Laplace trans
form of our unknown (unctton ts deterrmned Under general
mathematlcal conditions we can prove that our solution y 15essen
tlally" unique once tts Laplace transform is determined One
method of mversxon nsgn cn by contour Integration
1 c+uo
ya) = 53L Y(q)ed9. (114)
where the contour nsto the right of all poles of ?
ln any case, once 7(9) 15determmed, the determmatxon of y(t)
lSa detaxl which need not concern us Extenswe tables are aval!
able to and m mvertmg Laplace transforms 1nvolv1nga wide class
off's
To extend the theory we need only seek a fzmctzonalappropriate
to any admxssnbleh. thh the fundamental property
(q. by) = q(q.;v)h -N) (115)
[he followmg expressron formally ans ers our problem
__ F0 ?! 0k
(q. f),. _ 11m
-"-m.
M 0);. (116)
where 9 and f are such that
f(t)
{LTFU, q, 1.0)h =0 (117)
The proof that tlns 15our generalized Laplace transform ls gwen
us nJlmws
FU qu hf)
(q.hh= [1mm (118)
By denition
(h _Q)F(t!9109f)h 2f: F(lo,g.0,f)h "'-:O
(h q)F(t.q.1.0);. = o, F(lo,q,1,0),. = 1 (119)
(h - q)F(t.q. 0.110;. = hf. F(to. q.0.hf):. ==0.
and each of these systems has unique solutions. Now it can be
veried that
FU, gr 0; hf)l E hFU! g! ();!)h 'f(t)F(t9 g: 1; O)! (120)

because
hF(to, q, 0;f)h " f(to)F(to, q; 1;0)h E f(tO) f(to) = 0. (121)

and
(h 017th a, 0;f):. f(to)F(t. q, 1:0)h]
= 14:01- q)F(t, q. 0;f)h] - f(to)(h * (DFU.q. 1:0)1.
= h[f] O. (122)
Therefore,
(q; hf) = lim

[I__f
941,10
(): f(to)]
_ . Fa, q, om. . f(t) _ (123)
133.3
q F(t,q,1;0);. + EEE:Fu, q, 1; 0 N)
= q(q;f)h + 0 1%)
for 9 and f limited as in our hypotheses. Q. E. D.
Thus, the solution to the general nth order problem with pre
scribed boundary conditions is given by the inversion of
R__(_q) (q;
__f_);._
or formally
W) = .c-I [$$%]+ .2-1[Egg] (125)

Moreover, even where there is no convergence of , " can be
inverted as the mathematically inclined reader may work out
for himself. So long as f(t) leads to a unique solution of the rst
order operator equation, its behavior at innity is irrelevant.
Exercises:
1 Show that(q- m = 2 yf, = Yang. where to = o.

2. What 15(g; y)A? \Nhatq'ls (g; y)w?
3 Make and ven! y a table of transforms for E as follows
ya) ?(m
a'= F(ta10)a
:a :
( g l a ) :
t(! -1) (tn' n + 1)a' " ( g a

1 )"+x
,
F a 0 yz). 991:
81
>; 3'10)?! "' 1 'i) :(Q)zz(s

'y a) (q a);
a a
EyU)
4 Trace the relatlonshrp between the classmal and the generalized
Laplace transform method
MANY VARIABLE SYSTEMS

30 Let us return now to lmear systems of many vanables
(y: y: y) For the dynarmc system to be determmate, we
must have at least m operator equatxons These can, of course, be
of hlgh order, say ?: or less and can be wntten as
do uh"?! + au 12h"yz+ + 01 11h")?! +
+ au lmyn= jla)!
du 21hy1+ au My: + + 01 21h_y1+
n2mm=2tv
+0 y f() (126)
Go:th + Gomwyz + + 1 h""y1+

+ a mmyn: fm)!
Ol' as
P1101)?! + PnUyz + + PmUDm: 110)
P2100191+ PzzUyz + + P,(h)y= fa):
P1131003.
+ Pm2(h)y2+ + P(hb'm : fn)!
where P1300 are polynomials in h of the nth degree or less; or in
matrix notation as
ooh)! + alhHy + - - anlhy+ any = f @)
and
Py =f,
where (ao, - - -, a,.) are 1122matrices, y and f column matrices of 17:
elements. Of course, many of the elements in the (1smay consist
of zeros.
The maximum order of our system will be nm; but its actual
order may well fall short of this. By the actual order of the system
we mean the maximum number of initial conditions which can be
arbitrarily imposed on the system.12
As an exercise in matrix notation the reader may work out the
demonstration that by redenition of variables 13the system can
be written as a rst order one,
thY+A1Y = F, (127)
where
'I O O Ol
0 I 0 O
A0 = r
0 0 0 I
L0 0 0 ao
r0 I 0 0'
0 0 0 0 0
A1 = r F:
0 0 O I f
ran (Zn1 02 al
" It can be shown that the order of the system, $, is equal to the degree ih q of
the determinant
A(q) = laoq"+ a;q"'l + ' ' ' + anl
See R. A. Frazer, W. J. Dunn, and A. R. Collar, Elementary Matrices and Same
Applications to Dynamics and Derential Equatiom (Cambridge, 1938) for references.
See par. 5.4.
Frazer, Duncan. and Collar, Elementary Matrices. par. 55
Although our system has been converted to a rst order syster :,
1t15not yet m normal" form thh h Yexpressed exphCItly terms
of Y and functions of tlme
hY==AY+f (128)
If A0 nsnon smgular, that IS.1fxts determmant does not vamsh,
we can premultnply both srdes of (127) by A0 and Immedlately
convert to normal form
hY = (AO-IAM + (A.,IF) (129)
Srmxlarly m (126) 1fao xsnon smgular, our operators of hrghest
order can be solved for m terms of those of lower order by the act
of premultlphcatxon w1th do"
If au 15smgular our system 15not well specred One of three
Sltuatlons must then prevanl (1) there exxstpurely statzcal, algebrmc
relatzansths between some of the vanables By means of these
relatlons one or more vanables can be ehmmated, and the leading
matrix of the resultmg smaller system w111then be non smgular
(2) the relatlonshlps are not all mdependent so that a unique solu
tron does not emst Arbxtrary functlons of trme w111enter Into
the solutron (Example A system mth two Identlcal equatlons,
permlttmg one varlable to be chosen as an arbltrary functlon of
t1me) (3) the system 15mcons1stent
31 It 1s up to the economxst to present a well spec1ed dy
namlcal system In what follows we shall assume that the task
has been adequately performed so that whenever desu'able our
system can be wntten m the normal rst order form of order :
hy = Ay +f. (130)
w1th 1mt1al condltlons
3%) = b
As m prev1ous SeCtIODSwe shall nd our solutlon 1n two steps
rst. the solutlon of the reduced or homogeneous system
hy = Ay. (131)
3'00) = b
and then the solutron of the unreduced system
y(to) = 0
The sum of these two solutions will be the solution to our general
system (130).
32. Earlier we have seen that the solutions of the simple rst
order equation in one variable provided the building blocks for
the solutions of more complicated cases. We try, therefore, a
solution of the form
yl = CIFU:921:0: : CIF, g),
(133)
y! : CSF: Q, 1; ())/1: C,F(t, Q):

where F is as dened in earlier sections, and the 5s and g are un
known constants. After substituting into (131), we get by means
of the fact that M equals gF,
qch = A1161}?+ A1262F + ' ' + 14-13511?)
QCgF= A2161F + 44.22ch + ' + A2scaF:
. (134)
gCgF = 24.1ch + A2C2F+ ' ' + AncsF

The common factor F may be canceled from both sides of the
equation, leaving us with s algebraic equations which do not involve
time. If these equations can be satised, we shall have a solution
of the operator equation, although not necessarily the solution
with appropriate initial conditions.
Our linear equations involve known As,undetermined 6sand q.
Our equations are homogeneous in the 6'5 and may be written as
(Au _ g)01+ A1262+ ' ' + Alice : 0,
142.161+ (A22 g)62 + ' ' ' + A235: = 0: (135)
Aalcl + A1262+ ' ' + (Au _ QC. : 0

If all the 5sare zero, we have a trivial solution of the operator
equation. When can we nd a non-trivial solution? From the
theory of ordinary linear equations (Cramers rule, etc.), we know
422 FOUNDATIONS 0F ECONOVIC ANALYSIS
that a system of homogeneouslmear equatxons has a non trmal
solutlon 1f,and only li, 1ts determmant vamshes. m this case nf
An ? Au A..
Au A:: _ Q A:.
Mg) =
An An: A _ q
= n'-og.+ fig.l + + 'alq+ Tl = 0 (136)
For a q chosen at random the determmant ml] not vanish, and
ne shall heue no non tnvnal solution But q 15at our dlsposal, we
select xt to make the determmant \amsh The determmant ls a
polynomxal of the sth degree m q, and 15called the charactcnsc
equahon of the matnx (or secular equation) The roots of the
charactenstlc equatton must necessanly be : m number (countmg
multxphcmes) They are called characterxsnc or latent roots of
the matnx If e desrgnate them by (q. , g.), e can me
11(9) = (q: - q)(q: 9) (q. - a) ==0 (137)
It nsnot easy m fact to nd the charactenstlc roots of a matnx
From the demtxonof a determmant. the coelcnentsof the chame
tenstxc polynomial can be determmed as the sums of pnncnpal
mmors of a g en order (cf Mathematlcal Appendtx A, p 374). and
any of the customary methods for solvmg a polynomlal for tts
roots may then be employed Exen \\ 1th the best short cuts. the
computatxons are tedious Fortunately, the economxst does not
always require the quantitatne \alnes of the roots
Let us return to the problem of ndmg solutions to our reduced
operator system If e put a charactenstlc root. g.. mto the F m
(133), and take for our c's a non vanishing solutnon of (135), ne
shall have a 8011111108
whlch satnses our reduced system Note
that our selectlon olc's ISnot umque, doublmg each wxllalso nge
us a solutxon
A column of c's satlsfymg the homogeneous equatxons xscalled
a charactenshc teclor or latent vectoror sometxmes a modal column
or normal component Some of the charactenstxc roots may be
Frazer Duncan and Collar. ElementaryHolmes chaps w and v
complex; the characteristic vectors will then be complex. Of
course, conjugate complex magnitudes can always be combined so
as to give only real quantities in the nal answer.
If a root g,-is simple, then A(q,)
is not equal to zero, and the
rank of [A giI] is (s -1). Consequently, the latent vector
corresponding to q,-is unique exceptfor scale factor. We may dis
pose of this ambiguity by some normalization convention.
If all roots are distinct, the general solution can be simply
written. Where roots are repeated, certain complications occur.
However, these are not intrinsic and rarely concern the economist.
Later, I show how these difculties can be avoided. But for true
mathematical insight the reader must be referred to the theory of
elementary factors and divisors.
If the s roots are simple, and the 5 latent vectors have been
normalized, we have the following particular solutions of the re
duced equation (131):
CLIFO) gl); 61.2F(t) ga); ''' 61.0F(t1ga);
62.1170: Qi); 02.2170, 92); ' 62.317, Q.);
' ' (138)
C..1F(t, gl); 55.21%}!Q2); ' C._,F(t, gs)!
where F (t, g) is an abbreviation for F(t, q, 1; 0);,. Any linear com

bination of these is a solution, or
y,- = KICMFU. Qi) + KaCmFU, 92) + ' ' ' + K,c,-,.F(t, 9.).
(j =1,---,s) (139)
Note that the csand qsdepend only on the matrix and not on
the initial conditions. But the Ks must be determined by the s
initial conditions. Thus,
y1(0) = chm + Katina+ ' ' + KJ = bi.
yz(0) = K162.1 + K262.2 + ' ' ' + K|c2.a = 172.
(140)
y(0) : chaJ + K2553+ ' ' ' + K363_3

: bu
M. Bocher, Introduction to Higher Algebra (New York, 1933).
424 FOUNDATIONS op ECONOMIC ANALYSIS
01'
K = 6.
These equations are sufcient to solve for the Ks in terms of the
Us if the matrix c is non-singular. The reader may be referred to
any mathematical text for the proof that c is non-singular, and that
('1.46
= [9.5er (141)
In the case of distinct roots, the solution to the reduced system
(13!) can be written in compact form as "
y = c[F(t, q;).,]c"b. (142)
33 If b is the identity matrix, we have a fundamental set of
solutions This may be written formally as F(t, A). either as an
extension of Sylvesters Law of Latency mentioned in Mathe
matical Appendix A footnote 13, or from the denition
Fa. A) = m.(t)A. (143)

0
where the m's are dened by the formal identity
m. 9) = %mum. (144)
Because of the property
1;F = 9F, (145)
h Si mmq = {: m.(t)q. (146)

O 0
and we have the m

formala identities
,
hmfh = m;_.1(t). (r = 1, >, (147)
and hence
hF(t, A) 2 Am, A) (148)
Also
Fag, A) aa I. (149)
Formally. these above relations hold even if there are equal
roots. The formal solution given in this section is sometimes a
In Mathematiml Appendix A this fact is used in Connection With symmetc
. for wlucb : |: orthogonal, : e , c'!.. "
u qun Duncan,and Collar.ElementaryMamas. chap v.
very convenient one. For the differential equation system
Dy = Ay. (150)
F (t, A, 1; 0) takes the form
2
FD==I+A%+A2tZ!+H-. (151)
For the system
Ey = Ay, (152)
we have
F(t, A, 1; O); = A. (153)
However, for stability investigations these solutions are not very
convenient.
34. If instead of the roots all being simple, some are repeated,
we may write
Mg) = (a: q)"(92 - ) (qr 9). (154)
where (gl, -- -, q,) are the r distinct roots with respective multi
plicities (m1, - ~, m,).
As before, there are special solutions of the form, c,-F(t,qi), but
the c's are no longer unique except for scale factor; in fact, even if
the cs were uniquely dened, there would not be a sufcient num
ber of particular solutions to combine to get a general solution for
s arbitrary initial conditions.
Earlier, in the case of multiple roots we tried solutions of the
F(t. gs) 6F(t. qi) mFU, qi)
, and so we shall here.
form, aq , qz ) ' ': aqua1
Let R(q) be the adjugate of the q matrix,
PGI) = A - QI; (155)
that is, R(g) is the transpose of the matrix made up of the cofactors
of P(g).
By means of algebraic theory of no interest here, it could be
. dkR {
shown 13that from the columns of the matrix [ d? )] can be
selected vectors which when multiplied by the partial derivatives of
F will give a complete solution. However, these difculties can be
" Frazer, Duncan, and Collar, Elementary Matrices, par. 5.10.
sndestepped by the dev1ceof reducmg the s rst order equations
s dz'erentvanables to stngle sth order equatrons m each vanable "
Thus nf
by ==Ay. (156)
J'o) ="
then
hyUo) = Ab
hyuo) ==Ab
(157)
hy(to) = A "12
Because of the Cayley Hamllton Theorem that every matt-lxsans
es Its own charactenstxc equatlon or
A(A) E 0, (158)
we nd from (156) that
AGM = (zmh'+ uh"! + + my = MAM ==0 (159)
Hence, each vanable say y,(t), satxsesan sth order equation We
also know from (157) the values of [y,(to), hy,(to), , h'y,(to)]
Hence. we have a well determmed system whose solution can be
found by the methods of sectlons 24, 25, and 26 It ls clear then
that our nal soluttons take the form
31'091) "'1-1F(t, Q:)
y,) = 6, MF,Q!)+ cf __" + + 6, liml agmrl
fn,1,
+ C,21170,Q:)+ c, ___
35%:
___q_:) + + cmn,186953.192)
(160)
a r r ""I , '
+ C,"FU' %)+ C,___
ng q ) + + C:m1
a 6958f) v
(J = 1: ,S)
" The method of attack suggested here is not only theoretncally sxmplebut xsalso
computanonally optunal See P A Samuelson Efctent Computauon of the Latent
Vectors of a Matrix Proceedmgs of the Natzonal Academy of Science: XXIX (1943)
397
where the 52csare not all independent. All but 5 are determinable
from the matrix above. The remaining s depend on the : initial
conditions.
Thus, terms of the form Fk/q" may appear if there are multiple
roots. But they need not appear even if there are multiple roots.
That is, whatever the initial condition b, the coefcients c,may
be identically zero for (1 < k < mg). For example, in the matrix
A = E 35,3, (161)
3 is a repeated root of multiplicity :. F(t, 3) appears in the

nal solutions, but not F(t, 3)/q, or any higher derivatives.
The reader must again be referred to the theory of elementary
divisors for insight into this possibility.
As far as stability is concerned, except for borderline cases
(which we have excluded in our strong denition of stability), it is
clear that
lim F (t, 4) '= 0
implies "
lim = lim lim FU. 9 + Ag) - FU. q) =] im 0 = 0. (162)

[can q l-o Aqo-O AQ AqO
Therefore, it does not matter whether or not the partial derivatives

of F are suppressed.
35. We have now given the complete solution of the reduced
equation. By the same method as the last section we may indicate
the solution of the unreduced part.
by = Ay +f. (163)
yao) = O.
A borderline use on the boundary between stability and instability is provided

by a symmetrical conservative physil system. Here F(,iq)p = c" and 15neither
damped nor undamped. However, F/q = iteiqt is explosive, as ls well known from the
theory of resonance. But the autonomous reduced system y = Ay Wlllnot have cxploswe
solutions when there are multiple roots; for A = A, the theory of elementary dmsors assures
the vanishing of coefcients of all terms of the form (VN so that only an external forccaln
cause explosive resonance. [Added in 1951: L. ]. Savage (Journal of Political Economy, _ ,
1948, p. 202) has correctly pointed out the invalidity of this double-limit argument, _1us
triing With " = %(exp2!) X (D + 1) and F (t,q) = exp ( t + q exp 2!) for imaginary
q.
We easily see that
W = y.
hy = Ay +f.
hy ==Ahy + hf = A? + Af + hf.
hy ==Ay + Azf + Ahf + 11},
. (164)
ht-ly ==Acly + Aof+ Ase-Sh)?+ _, , + Aha-3f + hcZf.

hc : Acy + Ac-If + . , . + Abs-2f + hllf.
Multiplying the rst line by r., the second by 1r._1.the last by Wo,
and adding, we have
A(h)y = A(A)y + mof + mlhf + - - m._1h"f, (165)
where the rst term on the right-hand side vanishes by the Cayley
Hamilton theorem, and where the m's are linear combinations of
powers of A.
Each of the variables (yl, ' - -, y)must satisfy an unreduced
sth order equation. The righthanddriving functions seem to
depend upon hf, hf as well as f; and yet, we have nowhere made
the assumption that expressions of the form 11") need even be de
ned. If we write out a typical equation, say in y;, the paradox
will be dispelled. We have
A(h)yi = Rn(h)f. (166)

01'
_ man},
* A(h)
Because R1 is of degree (s -- 1) at most. it will be clear that

(by partial fraction expansion or otherwise) we can avoid all terms
of the form h"f. Only expressions of the type F(t,q. Inf); will
appear.
We have now only to specify the correct initial conditions.
Because
Ayo) = 0. (k = 0.1. )
our initial conditions should be
hyUo) = f (to).
h2y(to)= Aon) + tho).
(167)
h"y(to) = A2f(to)+ A3hf(to) +

+ AkaWo) + h'*2f(to).
We now have an unreduced sth order equation with prescribed
initial conditions which each variable must satisfy. Our problem
has therefore been reduced to that of earlier sections, which has
been completely solved.
Exercises:
l
1. Show that f lo c4("f(w)dw upon proper denition becomes the solu

tion of (132), when h is D.
:-1
2. Show that E AL1f(i)performs the same role for h equal to E, to = 0.
0
3. By Using the generalized Laplace transform 05305. dened earlier,

go from
hy = Ay +].
to you) = b
, _ Tii(Q) _ Ti,-(Q)
where [T.-;(q)/A(q)] equals [qI -ajl1 and consists of rational functions whose
numerator is of lower degree than denominator. Show that expressions of the
form h"f(t) do not enter into the solution in an essential way.
36. We have already seen in Section 21 of this Appendix that

it is necessary and sufcient for stability of a differential equation
system with constant coefcientsthat the real parts of the roots
of the characteristic polynomial should all be negative, i.e., the
roots must all be in the left-hand half of the complex plane. Other
wise we should not have the required dampening factors in our
exponential solutions.
Fortunately, it is not necessary to solve the characteristic equa
tion for its roots to determine whether these stability conditions
are satised. There exist known simple necessary and sufficient
conditions on the coefcrents of the polynomial which guarantee
stabrlrty These are known as the Routh Hurwrtz condxtlons"
Let our general polynomxalof the nth degree be wntten as
f(X) = aux" + an)!1 + + aX + a==0, (163)
where aocan always be regarded as posrtwe and walloften be taken
to be umty for convemence
If the &coectents are all real, mtus a faimhar fact that such
complex roots as occur w1llappear m conjugate palrs, (u + w) and
(u 10) If we write the polynormal as the product coil (X -X ;)
these complex roots wall combme to form real quadratlc factors
It follows, therefore, that every polynomlal can be wntten as the
product of quadratic factors and linear factors
f(x) "
G(X + mxXX + m,) (X + m,)
(X' + le + 61)(X2+ ng + C:) (169)
The lmear factors can be grouped mto quadratic factors except
when 1218 odd, m whxch case f can be wrrtten as a product of
quadratlc factors and at most one lmear factor
Now the roots of f cannot be stable,' unless the roots of each
hnear and quadratic factor are stable We must 1nvest1gatethen
these sxmplecases, whxchcan fortunately be treated exhaustwely
m a srrnple way The lmear factor
X+m1=0
clearly has a Routhlan root (one whose real part ls negatwe) only
1f Its coefcnents are of the same sngn For - m; 15lts root. and
m; must hence be posxtlve
A quadratlc factor of the form
X + bX + c = 0
has mots whrch can be wntten as
_ b:l:fm
X = 2
(170)
n E J Routh Stabdztyowaen State of Manon (London 1877) Adams Prue Essay

Dynamcs of a System of Rigid Bodies (6th ed London 1930) part II chap VI A
' mtz Ueber due Bedmgungen unter welchen eme GIe Chung nur Wurzeln nut
reellen Therlen besltzt Mathemmch: Annuler: XLVI (1895) 273284
If b2 - 4c is greater than or equal to zero, the roots are real and
the quadratic expression can be factored into the product of two
linear expressions, (X + ml)(X + m2). It is obviously necessary
then that the coefcients b equal to (ml + mg) and 6 equal to 217:
should both be positive; for the sums and products of positive
numbers must be positive.
We have so far proved only the necessity for the coefcients
to be positive in the real root case. Are these conditions sufcient
in the real root case? And what about the complex case? Leav
ing the latter aside, we easily deduce that positive coefcientsare
also sufcient to give negative roots. The rst term in the numer
ator of (170) is clearly negative. If we choose the negative sign
before the radical, we have one root which is clearly negative. But
the other root will be positive only if the second term outweighs the
rst, i.e., depending upon whether b2 is less than b2 41:. By
hypothesis, 6 is positive, and it followsthat the second term cannot
outweigh the rst. Thus, positive coeicientsare both necessary and
suicient for negative roots of a quadratic, when the roots are as
sumed to be real.
The complex case is even simpler. If we add the two roots,
(u + iv) and (u iv), the complex part cancels out, and b is
twice the sum of the real parts. If these are negative, then b must
be positive. In any case 6 is positive, or the roots would not be
complex, i.e., b2 4c is less than zero only if cis greater than zero.
Therefore, our coefcients must be positive if the roots are Routh
ian. This proves necessity. If both b and 6 are positive and the
roots are complex, it follows immediately from (170) that n must
be negative. This proves sufciency.
Thus, from our treatment of the real and complex cases, we can
enunciate the simple theorem: a second degreepolynomial has roots
whose real parts are both negative if, and only if, its coeicients are
both positive.
Immediately we can deduce certain necessary (but not sui
cient) conditions which must hold in the general case if the roots
are all to be Routhian. Each linear and quadratic factor must
have positive coefcients, and, therefore, all the coeicientsof our
polynomial must be positive. This necessary condition would also
be sufcient if we dealt only with real roots by virtue of Descartes
Rule of Signs discussed in every book on algebra.
But there is no reason in the general case to exclude complex
roots.2 What if some of the 6's are zero or negative. but in com
bining to form the a coefcients the positive coefcients alway
outweigh the negative ones? Then our 0 coefcients all being
positive would not be sufcient. A simple example establishes
this possibility.
(x + ne:2 .001X + 1) = x3
+ .999x= + .999X + 1 = o. (171)
Note that all the as are positive even though the complex roots
have real parts which are positive (= + .0005).
Now that we are warned. a number of possible modes of attack
are open to us. We may tackle the third and fourth degree cases
by themselves, knowing that but rarely will higher degree equations
be involved. Or we may deduce further, stricter necessary condi
tions, hoping that when we have found enough of them theywill
be sufcient.
Thus, in the third degree case we have
X: +01X2+02X+03 = (K+ m)(X+bX+c)
= X + (m + b))f2 + (c + bm)X + mc. (172)
We form the productnever mind why
11102 ana; = (m + b)(c + bm) mc(1)
= b(c + mb + m) + 0. (173)
If the roots are all Routhian, c, b, m are all positive, from which it
follows our expression ala, --
aaa, is positive.
We now have a new necessary condition. or four in all,
al > or
a; > O, (174)
a: > 0r
ala; aaa; > 0.
Are they sufcient? If we seek an example which satises them
and yet yields nonRouthian roots, we shall be disappointed. Or
so the author's experience suggests. But how can we be sure that
If the polynomial in question is the characteristic equation of a symmetric matrix.
roots must all be real, and the simpler analysis can be applied.
a more clever, a more zealous investigation will not produce such
an example? Only by a sufciency proof.
As a matter of fact, these four conditions will be later shown
to be both necessary and sufcient. But there is still one hitch
They are not all independent. The reader may verify that the
condition (12> Ocan be deduced from the other three. (Can any
one be deduced from the other three? Which can, and which
cannot?) It was the great achievement of Routh to have derived
an easily calculable set of independent necessary conditions.
As an introduction to what may be called Rouths test deter
minants, consider the 3 X 3 matrix
01 aa 0
ao (12 0 ' (175)
0 a; aa
Its three naturally ordered principal minors, A1, A2, A3, are, re
spectively, ai, 0201_ asao, de(aiz _ dado),and if they are all posi
tive, then the necessary and sufcient conditions for a cubic to be
Routhian are realized. This suggests a general rule for a poly
nomial of the nth degree. List its odd coefcients in a row,
treating all coefcients with subscripts greater than n as zero:
a1a3a5a7.. -. Let each of these elements head a column, whose
subscripts decrease one at each step. Follow the convention that
any negative subscript implies that the coefcient is zero. Thus
we have a square array of numbers, of which only the rst 7:rows
and columns need concern us.
Emil-"5;: ai: --- (a. > 0) (176)
The reader should write out the matrix in full for an 8th and 9th
degree polynormal, notmg the dxfference between an odd and
even 11
Routh's theorem states that 1!ts necessaryand su'iczentm order
for the real parts of all roots to be negatwe that the prmczpal mmors of
u: momx all be pos-mue, 1e ,
A], = 01 > 0,
Az 0102 " 000: > 0.
(177)
Ail1> 0'
An = anAnI > 0
The condrtlon that all the 0s are posntwe 1scontamed wnthm these
condntzons. but not vxce versa (except 1fall roots are real)
The ngorous proof of thlS theorem cannot be gwen here The
reader must be referred to Rouths Adams Prize Essay A brief
sketch of the reasonmg may, howet er, be 1nd1cated Let X be
any pomt 1nthe complexplane, and f(X ) be a mapping Into another
complex plane Consxder any closed contour C m the X plane
The number of roots of f (X) thhm thts conteur can be determmed
by an Important theorem of Cauchy,23Wthh IS essentlally topo
logical ln nature Traverse the contour C 1n the posxtrve sense
(counter clockwnse) and cons1der the mapped contour of f(X )
The number of roots 111C equals the number of times that f(X)
loops around the ongm , or 1fwe write f (X) as P + Q the number
of tnmesthat the ratxo P/ Q passes through zero from plus to mmus
m excessof the number of tlmes that the ratlo passes through zero
from rmnus to plus
Routh takes as hrs contour C the nght half complex plane. or
more ngorously a lumtlessly large semmcn'cleto the right of the
ongm and the 1magmary was The changes m Slgflof P/ Q on the
sem1c1rcleare easxly evaluated Hence, ll there are to be no roots
m the nght half plane, the number of changes m sxgnof P/Q on
the Imagmary ax1sare uniquely determmed. and the polynomxals
1300 and fg(X) must have all real roots whlch separate each
other, where
i f(zX) = MX) + AMX) (173)
See any textbook on complex vanables
By classical Sturmian theory this involves taking the greatest
common dVSOTOf MX) and f2(X) to get new functions f3(X),
f4(X), - - -. Finally, the conditions on these subsidiary functions
can be shown to be equivalent to the Routh-Hurwitz test deter
minants given in (177).
Numerically, these determinants can be easily evaluated by
various Gauss-Doolittle methods. As an exercise the reader may
verify that the conditions for a fourth degree equation are as
follows:
(11 > 0,
0102 _ 0003 > 0,
010203 ' 00032 _ 01204 > 0,
(179)
a4 > O.
It is worth noting that as we move continuously from a stable

set of coefcients to an unstable set that a,. or A,._1rst change sign;
also that a reverse numbering of the as must satisfy the stability
conditions.24
37. We have seen that for a difference equation system to be
stable it is not necessary that the real parts of the roots of the
characteristic equation be negative. Rather the roots must be
less than unity in absolute value. Instead of all lying in the left
half of the complex plane, they must all lie in the unit circle.
For rst and second degree equations it is not hard to work out
necessary and sufcient conditions by direct algebraic means. For
higher degree equations such methods are tortuous. Fortunately,
by a simple use of complex transformations we can reduce our
problem to the Routhian problem already given a complete
answer.25 The complex transformation
___
X =z+1
z1' X+1
Xi 0%)
converts the unit circle into the left half of the complex plane.
In P. A. Samuelson, "Conditions that the Roots of a Polynomial be less than
Unity in Absolute Value," Annals of Mathematical Statistics, XII (1941), 360364,a
numeril example is worked out and some further facts are noted. If a number of
A's vanish, then a further test must be made.
""P. A. Samuelson, Conditions that the Roots of a Polynomial be less than Unity
in Absolute Value," Annals of Mathematical Statistics, XII (1941), 360364.
436 FOUNDATIONS OF ECONOAIIC ANALYSIS
Also
Z ! oZ 1"+1Z+1"'Z1
f(X)_
"zf(__+_)____
a( +) (2L1))( ) ..
(181)
+ a(z + 1x2 n"-1 + o{z 1)
(Zl)
If the polynomial f(X) is to have no roots in the unit circle,
then the numerator of the right-hand side, regarded as a poly
nomial in Z, must satisfy the Routhian Conditions. If ne expand
out this polynomial in the rst degree case, we nd as necessary
and sufcient conditions that the root of X + a; = 0 be less than
unity in absolute value;
1 + a; > O,
1 ' (11 > 0.
or lagl<1 (182)
For the second degree polynomial, X + axX + a, = 0. the
reader can show that the stability region of the (a.. a.) plane is a
triangle dened by
1+al+2>03
1"az>0. (183)
lGx+ag>0,
It is to be noted that the conditions are always one greater in
number than the degree of the equation. This is true in the
general case of the nth degree equation.
[(X) = X + at)?"1 + -
. + a.X + a. = 0. (184)
We form the equation in Z,
doZn+ 1Znl
+ ' ' + nlZ+ an : as(z _ 1)'(Z + lyv
(185)
__ u-i & _ , azj
takJ k)z (_1) ( ] )Z
where
:: s!
= __' . $ !; I; 0
(t) (s:)!
0. 3<t (186)
ll
l! 0. t<0
DI FFERE.\'CE EQUATIOXS
Collecting terms we nd
(187)
" = 1 cu + de + ( 1)""an-1 +(-1)a..

In addition to the condition,
do > or
we have the Routhian conditions
A1>0,
42>O,
(188)
A,.>0,
where these are naturally ordered principal minors of
iii; ; s
do : DI (
OO on Que! NU (189)
~9I9n
.
h
The reader may verify that for a.cubic equation,

(o=1+a1+az+a
l =3+alaz3GJ,
52=3 'axaz'l'3as,
; = 1 a1+a, as.
38 Formally, our stabxhty cnterna are complete But for
most theoretxcal purposes numcrlcal values of the vanous coe
ments are not at hand lt ould be hnghlydesxrable to be able to
mfer from the quahtatwe propertles of our dynamical matnces
even before thelr charactenstlc determlnants have been (labon
ously) expanded out whether or not they are stable
UnfortUnately, thrs xsrarely posmble Often. however, certain
very general sufcxent condntlons can be 1nd1cated to guarantee
that a d1erent1aleqnat10n xsstableor as I shall term at Routhxan
or to mdxcate that a dnflerenceequation rs stable mth modull less
than umtyor as I shall term 1t Tlnbergeman m honor of Pro
lessor Tmbergens outstanding work thh dxflerenceequations m
economics
The followmg theorems are lasted more or less wzthout proof
Theorem I :: = ax 15stable, ll the mam-< a 15symmctncal and
negatwe demte le , a. 15a Routhtan mattvc (Sec Mathemattcal
Appendlx A, page 19, passzm for thls classncal fact )
TheoremZ If () xs quas: denite, so that alzaIS nega
tlve demte, then a xs Routhlan (Hunt %(X 'X ) = X'aX
= X' (a-i) X < 0 Therefore, hm

lon x.(t) = 0 and roots must
be damped )
Theorem3 If a'a I xsnegatlve demte, thena 15Tmbergentan .
1e . X ,...; = aan stable, and roots of Ia -RI] = are all less than
umty m absolute value (Hmt Xg+1'X+1= X {(du I)X.
= AO: x.) < 0 by hypothesxs Therefore, lug X. = 0)
Theorem4 (Metzler) Il a IS anksxan as dened m chapter
VI , thh all tts o' dxagonal elements posxtwe, then xtxsnecessanly
Tmbergeman
Theorem5 (Metzler) If au 0 and f: a., < 1, then (a I) rs
Jul
Hickman, and hence (by Theorem 4) nsTmbergenlan
Theorem 6 (Mosak) If a 15chkSIan, and ll the o dlagonal
elements of a are all posxtwe. then all of the elements of 01 are
egatzve
Theorem 7 (Oppenheim-Yntema): If all the symbols are posi
tive in
A = (1'+ Z aii)6ii ati

bk: I (Wi.- + 2: bien)5L1] ,
then the pattern of signs in A1 is

+ + _ _
INDEX
ACCELERATION PRINCIPLE AND MUL BUSCIIEGUENNCE, S., 155
TIPLIER,interaction between. 341n. BUSINESS CYCLE, 350
"ACQUIRED CHARACTERISTICS", 211 billiard table" theories of, 340
ADVERTISING EXPENSE, 41 exogenous and endogenous the
ALExANDER, S. S., 155 ories Oi, 3355.
ALLEN, R. G. D., 125n., 146, 180n., nature of the, 3355.
196 non-linearity and amplitude of,
See also Hicks and Allen 3375.
"ALL OR NONE PHENOMENA,210, 241
ANALOGIES, BIOLOGICAL, 311 CANNAN, E., 223n.
ANALYTIC FUNCTIONS AND HIGHER CAPITALIZATION OF RESIDUAL PROF
DERIVATIVES, 525. ITS, 87
"ANNALS os MATHEMATICAL STA CARATIIEODORY,C., 363n.
TISTICS," 343n. CARDINALUTILITY, 915., 13Sn., 1725.
ANTI-TRUST, 203 and welfare economics, 196,226fl'.
ASYMMETRY AND I NSTABILITV, 187 CASSEL, G., 312m.
See also Symmetry CASSELs, J. M., 210D.
CAUCIIY THEOREM ON ROOTS OF AN
BARONE,E., 2125., 213, 245, 252 EQUATION, 434
BASTIAT,F., 211n. CAUSAL SYSTEMS, 285m, 315, 3175.
BATCIIELDER,P. M., 380n. CHAMBERLIN,E. H., 210n.
BENEFIT mon; FREE TRADE, 204 CHAMPERNOWNE,D. G., 125n.
BENNION,E. G., 34m. . CHANDRASEKHAR, S., 31711.
BENTHAM,J., 90 CHARACTERISTIC,equation, 271, 299,
BERGSON (BURK), A., vii, 219 223, 371n., 422; mathematical theory of,
225, 252 3735.
BERNOULLI'S LAW OF UTILITY, 227n. roots, 271, 299, 349, 371, 422
BIOLOGICAL ANALOGIES,311 vectors, 372, 422
BIRKHOFF,G. D., 54n., 262n., 300n., CHARACTERISTIC EXPONENTS OR MUL
319n., 334n., 339n. TIPLIERS, 339
BISHOP, R. L., 189n., 198 CLARK,I. B., 35, 211n., 313
BocneR, M., 137n., 423n. CLARK,J. M., 242, 313n.
BHM-BAWERK, E., 188, 23311. Coawsn CYCLE PRENOMENON, 22,
BOOLE, G., 380n. 265, 321, 339n., 3905.
BORDERED DETERMINANT OR MATRIX, COLLAR,A. R. See Frazer
37, 62, 101, 132, 167, 365, 3775. COMPENSATED CHANGE, 103
Bomxwrcz, L., 146 under rationing, 168
BOUNDARY OR CORNER MINIMA, 695. COMPLEMENTARITY, 1835.
BOWLEY, A. L. 146, 150 COMPOSITE COMMODITIES, 141, 144
BROMWICH-WAGNER CONTOUR INTE CONSTRAINEDMAXIMA AND Mumu,
GRALS, 398 60, 350, 3625.
BROWNIAN MOVEMENT, 268, 317D. See also Bordered Determinant
BUDGET EQUATION, 98, 166 CONSUMER'S SURPLUS, 1955., 206
BURK, A. See Bergson, A. six versions 0!, 199
441
442 I'OUNDAI'IONS Ul ECONOMIC ANALYSIS
CONsumno-I Fuxcnox. 276 ECONOMICS,boundaries of 319
CONTRACTLocus, 238 251 EDGEWoaI-E, F Y, 90, 92, 93. 146,
generahzcd. 214, 233 183, 206 225. 22611 251
Com: LAws. 250 ! LASTICITYConrrxczsws, 105 12511,
CORRESPOLDEhCE PRINCIPIE, 5 258, 188n
263, 284, 350 d1mens10nsof, 12
COST AM: Pnonucnor. THEonv or FLEEEMARV FACTORSAND Dmsous
THE FIRM, 57H , 350 37311 . 423. 427
Comme-:, A A , 141 ELLIS. H S , 11711
Count, L , 5511 EwoocEJous Tusoaxas See Bum
CEIrEnION FOR DISPLACEMENT or ncss Cycle
EQUILIBRIUM 3911 ENGEL's 1.111.113,193
EQUILIBRIA, MULTIPLE. 49, 240
DAVIDSON, D . 328::
Se: also Determmateness
DAVIS. H T , 34211 EQUILIBRIUM, 8
DECREASINGCosrs. 210, 239, 242,312 absence of, 26111
DEDUCTIVB REASOMNG 12 and varymg tune penods, 3305
DEFICIT FINANCING, 354 external condmons, 81, 88
DEMAND.{or a group of commodmes,
recedIng, 326
See also Determmateness
141
ETHICAL ASSUMPTIONS, 2033
for consumers good-9.965 EULERs THEOREMou HouocENEous
for factors of productIon 45 55. FUNCIIONS, 68, 83, 105
S9
EVOLUTION AND ORGAN: Gnowm,
DESIGN. TELEOLOGICAL, 203 312
DETERMINANTSee Matnx Eanueron or THEPnoovcr, 83. 88
DEIERMINANIAL EQUATION Sec ExocENous THEORIES Sec Bus:
Charactenstnc equanon ness Cycle
DEIERMINAIENESS or EQU1L1nmm1, EXTERNAL ECONOMIES 011 DISECONO
6211 , 75H , 145, 351 mms, 208
DIFFERENCE EQUATIONS. 265, 282, 111consumpuon, 224
287, 30211, 314, 3805
types of stabllxty pomts, 305 FACTOR or PRODUCTION. 84
DIFFERENTIAL EQUATIONS, 263 ,
FELLER, W., 34911
287, 288, 299, 314. 3806 FINIrE CHANGES,analysxs of. 46, 109
DIMINISEING RETURNS, 12, 23, 62 FINII'E DIFFERENCES. 381'
DIRAC FUNCTION, 286, 333
FISHER. I , 9311 . 95, 13911 . 146 173n ,
Dlsconrmvmas, 705 , 80 17411 , 183, 22811
DIscanNA'IINc MONOPOLY AND
Poucet: Mortes, 32511, 335, 341
PRICE, 42, 247 FRANCE. ANATOLE, 22511
varsxA. F , 15011 FRAzEn. R A , DUNCANANDConua,
Doerca, G . 398m 41911 . 42211 , 42411 , 42511
Dnascn, F , 268 FREE ENTRY, 87. 88
DUNCAN,W ]. Sec Frazer FREE TRADE, 204
Dunn-r MARSHALL.21811 FRICKEY,E. 146
DURAND, D. 12411 FRIEDMAN, M , 18011 . 181
DYNAMICS, 350 ansca, R. 811, 146, 174n , 196, 218,
and statics, 5:): categories of, 315 228 . 26111 , 28411 , 28511 , 31411 ,
comparatwe, 351. 355 31711 . 32811 , 342n
denmons of. 284m. Sll deninon of normal ethbnum
value. 3275.
INDEX 443
FRISCH-HOLME MIXED DIFFERENCE 62n.,
HOTELLING, H., 55n., 78n.,
DIFFERENTIAL EQUATION, 309 218n., 231, 252, 351
FUNCTIONAL EQUATIONS, 260, 286, Hunwnz, A., 430n.
3085, 3805.
IM KLEINEN" MAXIMUM, 206, 241
GAUSSIAN. See Normal IMPLICIT FUNCTION THEOREM, 48,
GENERAL EQUILIBRIUM, 8 145, 259m.
GEORGESCU-ROEGEN,N., 13911., 18211., INDETERMINACY IN PUREST COMPE
351 TITION, 785.
GIBBS, J. W., 21n., 70n. INDEX NUMBER, best limits on, 149
GIFFEN'S PARADOX, 115 economic, 111, 1465.
and rationing, 169 Laspeyre formula, 155, 162
GILBOY, E. W., 281n. non-existence of perfect quantity
GOSSEN, H. H., 93 indicator, 154
GRESHAM'S LAW, 170 of price, 156
of quantity, 160
HAAVELMO, 'I., 343n. Paache formula, 162
HABERLER, G., 146, 196 INDIFFERENCECURVE OR Locus, 945.
HANCOCK,H., 370n., 371n. INDIVIDUALISTIC HYPOTHESIS, 223
HANDICAP SETTING" and Optimal INEQUALITY, 211, 225
Taxation, 247 INFANT INDUSTRY, 253
HANSEN,A. H., 340n., 341n. INFERIOR GOOD, 169, 201n.
HARMONIC DIAL, 342n. INFINITE SUBSTITUTES IN WELFARE
HARROD,R. F., 340n. FUNCTION, 222
HART, A. G., 143n. INITIAL CONDITIONS, change in, 352
HAWTREY, R., 340 INSTITUTIONAL ADVANTAGE, 87
HAYEK, F. A., 214n. INTEGRABILITY CONDITIONS, 54, 95,
HEAVISIDE,0., 4015. 107n., 179
HEAVISIDE-CAUCHY OPERATIONAL INTEGRAL EQUATIONS,286, 309, 314
CALCULUS, 3975. INTEGRATING FACTOR, 95
HENDERSON, A., 189n. INTEREST, lack of necessity for uni
HENDERSON,L. J., 5n., 312 form real rate of, 233n.
HERMITIANMATRIx, 271, 3715. INTERNATIONAL TRADE, 204
HERTz-HERGLOTz SOLUTION OF IN INTERPERSONALadditions of utility,
TEGRAL EQUATIONS, 309
206 .
HESSIAN DETERMINANT 0R MATRIX, comparisons, 91, 205, 220, 245,
31, 49, 52, 67, 376n. 249. See also Welfare Eco
nomics
HICKS, J. R., vii, 18n., 86, 122, 123,
130, 133n., 143, 187, 189n., 208n., Optimal welfare conditions, 2435.
219n., 252n., 269, 276n., 317n.
HICKS AND ALLEN, 95, 96, 139, 184, JACOBIAN DETERMINANT OR MATRIx,
351 1275.
HICKSIAN MATRIX, 133, 139, 141, JEVONS. W. S., 90, 93, 146, 205
272, 438 JOHNSON, W. E., 95, 115, 351
HICKSIAN STABILITY AND TRUE DY
NAMIC STABILITY, 2735. KAHN, R. F., 210n., 219n., 260n.
HISTORICAL SYSTEM, 3165. KALDOR,N., 219n., 252n.,260n.
HOMOGENEITYof order one, 84n. KALECKI, M., 167n., 337m.
of order zero, 68, 105. See also KARMAN, T. VON, 339n.
Euler's Theorem KEYNES, J. M., 118, 122, 146
KSYLESIIN SYSTEM,28, 276, 353 Manama. PRODUCTIVITY,
S8, 83
Kucm. F H , 123, 19611. 20811, net. 75
21511, 223, 226, 311 of the last dollar, 60, 88
Hows, A , 146, 155. 196 MARGINAL PROPENSITYTo Consume,
Kownnmrr Love WAVES.
340 and stablhty, 27917
111111511111,UTILITY or INCOME. 99,
LAGRANGEAN Mumu-1.113115. 37. 6011 . 104, 135, 190
98. 131. 166. 192. 231 constancy of, 17211, 1891f
See also Puces MARGINAL UTILITY or MONEY. 190
Luca, 0, 12711. 17311 21911 23211, Manon? CHAIN,317n
252, 274n, 27611, 35411 239 MARSCIIAK. J . 12111
LAPLACE TRANSFORM, 398 MARSHALL. A . 6. 1811 . 2111 . 3511 , 79.
generahzed for any operator, 93n . 141, 146. 173. 19411, 197, 198.
41511 206, 207, 212 242, 246. 26411, 311
LASPEYRE See Index Number MATHEMATICALTHEORY. of character
LATENT See Charactensnc 1st1cequanon, 373E
LAUIHARDT, W , 205 of d1erence equatlons, 33011
LE Caux-man PRINCIPLE. 36. 3811 , of {unctIonal equatlons, 38011
81, 168 of general Operator equattoma,
LE CORBEILLER, Ph . 34011 39611
1.110311an Tamseonuuxor, 13811 of generahzed Laplace transform.
LEONTIEF, W , vu, 2911, 130, 146, 517
162a , 196, 33911 , 390n of maxlma, 35711
LER'R, A P. 146. 196, 217, 21811. of maxIma under constraints.
245, 246. 252 36211 (See also Bordered
Lassa, C E V. 12211 determmant)
LEvmsox, N , 33911 of quadrauc forms. 36517 (Se:
LEWIS D C .)1' Sec BIrkhoH,G D also Transformauon)
LIMITATIONAL FACTORS. 71 MATRIX. rank of, 132, 36211
szamu ANDNONLmzamror tranSpose" of, 132
SYSTEMS. 284. 288, 332, 39111
Se: also Bordered determmanl.
pruuco-rr. B E . 23211 Hermman, Hessmn, H1cks1an.
LIQUIDI'IY Pal-:anszsce, 123, 276 Jacobxan, Quasz demte, Rou
LOCAL [ADEPEVDEMCECommons th1an, Tmbergenlan,
180 Symmetnc and asymmetnc
MAXIMIzATION of prot. 7617
LOGISTIC LAW, 291
of ordmal uuhty. 97H
LOTLA, A J . 292m . 30911 . 32311
MAXIMUM CONDITIONS, 29. 48, 258.
3571? , 362E
LUZMP
47 SUM TAXES on Boumms. 245,
See also Quadratzc forms
Luwnzxc. E , 38911 1.leva PROBLEM.1517. 21, 98
LUrz, F A . 12411
convertanhty Into a, 52H
MAXIMUMUNDER Constanta-3,3611 .
Mamans, T R , 20311 98, 131, 350
MALTIIUSIAN AND OPTIMUM POPULA See also Bordered determmaut
11011Tuzomes. 296 Mama, J B. 27611
MANN, H B . 343a MEANINGFUL 131150115115, 3, 4,
MARGINAL COST. 65, 24011 4611, 84, 91, 107, 172, 221, 257. 284,
Anon. EFFICIENCY op CAPITAL. 350
three sources of, 21
INDEX 445
MENGER, K., 90 PLANNING, Optimum economic,
METZLER, L. A., 273n., 438 203ff.
MILL, J. S., 313 POSSIBILITY FUNCTION OF SOCIETY,
MlLNE-THOMSON, L. M., 380D. 236, 244, 250, 251n.
MINIMUM AVERAGE COST, 86, 88 PRICE DISCRIMINATION, 197
MINIMUM TOTAL COST, 58, 59, 66, 73, PRICES As LAGRANGEAN MULTIPLIERS,
88 216, 231, 235, 247
MITCHELL, W. C., 341n. PRODUCERS' SURPLUS, 208
MONEY,demand for, 117'. PRODUCTION FUNCTION, 57, 230
MONOPOLISTIC COMPETITION, 210 homogeneous, 83
MOORE, H. L., 315n., 321 See also Euler's Theorem
MOSAK,J., 273n., 438 See also Homogeneity
MOULTON,F. R., 388n., 408n. PROFITS, 8Sff.
MOVING EQUILIBRIUM, problem of PURE COMPETITION, 78, 82
denition Of, 321'. as a norm, 204
stability of, 329
MUTUAL INTERDEPENDENCE AND QUADRATIC FORM, denite, 30n., 55,
CAUSATION, 9 68, 77, 126, 140, 366. (See also
Maximum conditions)
NEISSER, H., 167n., 170n. denite under constraint, 62,
NET AND GROSS REPRODUCTION 376R. (See also Bordered Deter
RATES, 299 minant)
NEUMAN, J. VON, 170n. denition of, 30, 365K.
NORLUND,N. E., 380n. Semi-denite, 113, 3665., 379
NONCOMPETING GROUPS, 211 QUALITATIVE RELATIONS, a calculus
NORMAL OR GAUSSIAN DISTRIBUTION of, 23H.
OF PROBABILITIES, 343, 345, 349 QUALITATIVE RESTRICTIONS, 20
NUMERAIRE, 106, 117, 187, 190 QUASI-DEEINITE, property of a ma
trix, 140, 438
OPERATIONAL SIGNIFICANCE. See
Meaningful Theorems RATIONAL BEHAVIOR, 90, 98
OPPENIIEIM-YNTEMA MATRIX, 439 RATIONING, 167m.
OPPORTUNITY COST, 88n., 234n. point," 165
ORDINAL PREFERENCE, 22, 91, 945., simple, 163
1725., 221'. RAYLEIGH, LORD, 194
REAL COST, 211, 234m.
PAACIIE. See Index Number RECIPROCAL RELATIONS, 89, 103, 107
PALEY, W., 226 See also Integrability Conditions
PARETO, V., 93, 96, 139n., 141, 183, REDER, M, W., 143n.
191, 212n., 213, 236, 245, 249, 252, REGULAR MAXIMUM OR MINIMUM,
351 30, 55, 358E.
PERIODOGRAM AND FOURIER ANAL RELAXATION OSCILLATIONS and non
YSIS, 342 linear systems, 338'.
PERIOD OF PRODUCTION, 1886. RENT, 87, 208
PERPETUAL STEw, 188n. RESIDUAL THEORIES, 87
PERSONS, W., 146 RICARDO, D., 35 .
PFAFFIAN, 95, 139 ROBBINS,L., 219n., 220, 223, 249, 313
PICARD,E., 289n., 300n., 334n. ROBERTSON,D. H., 208n.
PIGOU, A. C., 146, 182n., 196n., 207, ROBERTSONIAN IDENTITIES, 278
215, 245, 246, 249 ROBIN HOOD, 206
446 FOUNDA TIONS OF ECONOMIC ANAL YSIS
ROBINSON.
G See thttaker Suu-ms, Connu-twa 8. 235811r
. 350
ROBINSON, JOAN, 7Sn STIGLEII,G J . 25m
ROUTE. E J . 43011 STOCHASTICAL SYSTEM 316E
ROUTE Hvxwrrz CONDITIONS FOR non Imear. 34411
STABLE ROOTS. 430 Sur111, W F , 23411
ROUTEIANMATnIx, 431. 438 Supzarosrrxon THEOREMAND LINE
SAMUELSON, P A , 11111. 12311 . 13911 , nm. 352
18911 , 21911 , 23411 , 25211 , 27311 , SUPPLY mn DEMAND 257
311n , 34111 , 38811 . 40811 , 40911 , SUPPLY CURVE, 41, 51, 7711, 260, 264
426n 43511 Swann. A R , 9111
SAVAGE, L J , 42711 SYLVESTER'S LAW of Inema, 36911
SCAncrr'l, natural and contnved, 87 of latency, 37511, 424
SanEsmcEn, K . 17011 SYMMETRY AND ASYMMBTRY, 14011 .
SCHULTZ, H , 104n , 12811 , 15511 , 141, 187, 272n , 438
18411 811111131111:or PERSONS. 224
Scnwrmxn, J A , vn, 117, 312n ,
31611 , 32011 , 34111
TAX, dIrect vs IndIrect, 197
Scrrovsxv. T . 16711, 219n , 25111
SECONDARYErrmmml CONDITIONS doctrIne of equal or m1n1mum
sacnce, 226, 227n
See MaxImum and Quadrauc Form formula, curvature" and nsks.
SECULAR EQUATION, See Character 248a
Isuc Equatwn problem. 14, 39
Sem- EVIDENITau-ms, 84
TEMPORARY NATIONAL ECONOMIC
SELF GENERATINGCYCLEand urcular
COMMITTEE(TNEC), 203
reasomng, 336
THERMODYNAMICS. 2111 ,36
SENIOR, N . 211
Smcwxcx, H , 90, 226 THIIFIINESSand
3 3 capital formanon
SLUTSXY, E , 4n , 102, 10311 . 115. 167,
18011, 182, 18411. 31711, 343, 351 TH'NEN, J H 211
TINBERGBN, J , 8n , 34011
5111111,ADAII, 203, 313
TINEEIIGENIAN MATRIX 438
Sun-11,Oman K , 33911
SOCIAL WELFARE FUNCTION, 2198 , TRANSFORMATION,cogredlent. 133
2265 . 248 " congruent, 133, 138
enght assumptxons about. 22111 contragredxent, 137 274
SOCIOLOGY, 90
curve of socty, 230, 234
STABILITYcondmons, 5, 25811 , 350 loganthmlc, 125
of commodItIes and pnCes, 135
(See also Correspondence Pnncnple)
and maxnnum condItIons 301 of dependent vanable, 133
danger of assummg non dampen of mdependent vanables, 129
mg, 268, 336 of tune, 318
dxerent concepts of, 26111, 272, of utlhty. 94, 175, 37611
290, 300, 312, 53311 TWEIIEBETTE-n"Psorosmoxs,
lugher order, 294 250. 251
of multiple markets, 269
of smgle market, 18, 263 UNCERTAINTY. 117
Walrasxan and Marshalhan, 26411 U SHAPED COST CURVES 85
STACKELEEIIG,H von, 1701: UTILITARIANXSM, 206
SrAEHLE, H . 146 196 UTILITY.evolutlon of concept, 905
STAMP, Su- ] , 22311 Independence of. 1741?
vs STATIONARY,313 Index, 94
INDEX 447
invariance of, 95, 99, 104, 183, optimum conditions between per
193. (See also Transformation) sons, 245.; in exchange, 212,
mathematical aspects of, 925. 218, 236'.; in production, 217,
See also Marginal utility of in 218, 230'.; of degree of product
come differentiation, 197, 232m, 241:
of saving, 25311.
VALUE JUDGMEN-rs, 220 WELFARE FUNCTION. See Social
VAN DER POL, B., 339n. welfare function
VEBLEN, T., 119, 224 WESTERN MAN, 226
VERHULST-PEARL-REED. See Log WHITTAKER,E. T., 381n.
istic Law WICKSELL, K., 86, 146, 204n., 205n.,
VINER, J., 34, 35, 66n., 197n., 208, 206, 212, 215n., 32811.
219n., 260n., 312n. WIENER, N., 34211.
le MEDICATRIX NATURAE," 312 WILSON, E. B., vii, 81
WINTNER, A., 13811.
WALD, A., 146, 150, 170n., 196, 34311. WOLD, H., 3171134211.
WALRAS, L., 8, 79, 86, 90, 93n., 118, WONG, Y. K., 34, 6611.
141, 204, 205n., 260, 264, 351 WONGVINER-HARROD ENVELOPE
three theorems Of, 87 THEOREM, 243
WELFARE ECONOMICS,vii, 52, 173,
20362, 220 YNTEMA, T. O., 439
and cardinal utility, 19611. YOUNG, ALLYN, 146, 208n.
and rationing, 171 YULE, G. U., 31711.
modications introduced by dy
namical considerations, 253 ZONE OF INDIFFERENCE AND IN
new and old, 212, 249 TERNAL MARGIN, 212

Paul Anthony Samuelson-Foundations of Economic Analysis-Cambridge, Harvard University Press, (1947)

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Paul Anthony Samuelson-Foundations of Economic Analysis-Cambridge, Harvard University Press, (1947)

Hochgeladen von

Copyright:

Verfügbare Formate

HARVARD ECONOMIC STUDIES

THE STUDIES IN THIS SERIES ARE PUBLISHED BY THE DEPARTMENT OF

UNIVERSITY OF IODHPUR UBRARE

NINTH PRINTING. I971

DISTIIBUTED i GREAT BRITAH DY

UNIVERSITY F Igor-twa I_JBRARE

Acknowledgment IS made to the editors of Economemca and

and a nomogram (p. 171)

Both of thesenomograms can be used onlyafter2x2, 2y2, 2xy have been

FoundationsofEconomicAnalysis.Paul AnthonySamuelson(Professorof Eco-

HIS is a very important contribution in the fieldof pure economicswhich

This content downloaded from 192.12.88.143 on Wed, 3 Sep 2014 08:19:34 AM

The firstpartofthe bookis concernedwitheconomicstatics.It also in-

This content downloaded from 192.12.88.143 on Wed, 3 Sep 2014 08:19:34 AM

This content downloaded from 192.12.88.143 on Wed, 3 Sep 2014 08:19:34 AM

II. EQUILIBRIUM SYSTEMS AND COMPARATIVE STATICS

The Stability of a Stationary Position which

This in brief is the method of comparalwc statics meaning by

posszble to speak ol'c msataon attnbu table to each except m respect

"'(xl! val all ! a) = ou

and the corresponding set of unknowns

The rst condition merely states that at a maximumthe tangential

For each value ol t, there W111

%,;ww) C(x)1(%%)= 1. (12)

$$ Ex1>(x) C<x):| < 0 (14)

(Es-:5) < 0 or g'(t) < 0 (15)

Let us consider as another example a market for a good or

(%;)o s' > o (23)

Before going on in the next section of our work to indicate how

b The method of camparatwe slam: consnstsof the study of

THE THEORY OF MAXIMIZING BEHAVIOR

\Ioreover, 1t ISpossxble to denve operattonally meamngful resta:

:l: i :1: ... +

NlAXIMUM CONDITIONS or EQUILIBRIUM

] f: ,,Ohh, 0' (10)

% )f,,,h.h, < o, (11)

2::fziz,.%;: fxiakO, (i = 1, . . 3.72) (13)

II = ' = Ifzijol (16)

far-210 "' fz.zo

:iply the 15thequatioan

holding for any h's. In particular, for

This may be veried by successive differentiation to obtain

For the rest of this chapter unless otherwise indicated I shall

DISPLACEMENT OF QUANTITY MAXIMIZED

53:=f(x. x,. a) = 0 (z = 1 n) (31)

prots (etc) nt does so most when the unknowns are optnmall

Auxruamf ConsTRAINTs AND THE GENERALIZLD

If the equnhbrxum of a system ISdetermmed by ewtremum con

mx,- -x9) = 0, (a = 1, r) (39)

i g(xi _ xJ'o) : 0! (B 11 ' ': T)

where the Xsare Lagrangean undetermined multipliers. _

% J is negative semi-definite. (Cf. Mathematical Appendix A

by a well-known theorem on determinants (Jacobi). __

While the change in an x with respect to its own parameter is

To a rst approumatron. the change m prots ts unaectedby

gg- = 0 = Diab) + D''(i>)

llns must of course determine the same solution as given m (66)

Now the computation of (ml.-,while laborious, is nevertheless

g=gi(a)<o aab (88)

Ax = gel) g<a>= j: g'<a)da = g'Aa. (91)

where (xl, -, x) take on any values. For brevity this may be

z = 0(x1, - . ,x,.) )am. (114)

6(x) 213013303; 9(x) me,, (115)

6(x) ' Z:jagx,- 0(x1) d.-cx}. (116)